The Customer Loyalty Solution

Transcription

Praise for
The Customer Loyalty Solution
“Arthur blows past the CRM hype and lays out the best of database
marketing, presenting case study after case study of how to do it
right (and sometimes not so right!). His integration of current
marketing strategies, database marketing techniques, and how the
Internet really helps database marketers provides new insights that
everyone will learn from. Required reading!”
—Eric Webster, Assistant Vice President, Customer Marketing,
State Farm Insurance
“Provocative, stimulating, interesting, and loaded with case studies
from dozens of companies, Arthur’s book should be the bible for
anyone doing cutting-edge database marketing today.”
—Mike Brostoff, Chairman, CSC Advanced Database Solutions
“An interesting and entertaining read which provides practical
insights into the evolving world of Database Marketing/CRM.
Arthur's ability to bring focus to the day-to-day challenges (and
misconceptions) of the discipline makes the book a great resource
for the experienced database marketer as well as the novice
practitioner.”
—Robert Burgess, Group Manager-Customer Relationship Management,
Verizon Information Services
“A lifetime's worth of experience and understanding packed into a
fun and easy read for the novice and the expert. Add the excellent
examples of companies that have been able to ‘make it work,’ and you
have an invaluable resource for anyone seeking to succeed in their
1-to-1 or database marketing efforts. A must have for today's
marketing manager.”
—Kay M. Madati, Relationship Marketing Manager,
BMW of North America, LLC
Also by Arthur Middleton Hughes
The American Economy, 1968
The American Economy, 2d ed., 1969
The Complete Database Marketer, 1991
Strategic Database Marketing, 1994
The Complete Database Marketer, 2d ed., 1996
Don’t Blame Little Arthur; Blame the Damn Fool Who Entrusted Him
with the Eggs, 1999
Strategic Database Marketing, 2d ed., 2000
The
Customer
Loyalty
Solution
What Works (and What Doesn’t)
in Customer Loyalty Programs
Arthur Middleton Hughes
McGraw-Hill
New York Chicago San Francisco Lisbon London
Madrid Mexico City Milan New Delhi San Juan
Seoul Singapore Sydney Toronto
Copyright © 2003 by Arthur Middleton Hughes. All rights reserved. Manufactured in the
United States of America. Except as permitted under the United States Copyright Act of
1976, no part of this publication may be reproduced or distributed in any form or by any
means, or stored in a database or retrieval system, without the prior written permission of the
publisher.
0-07-142904-2
The material in this eBook also appears in the print version of this title: 0-07-136366-1
All trademarks are trademarks of their respective owners. Rather than put a trademark
symbol after every occurrence of a trademarked name, we use names in an editorial fashion only, and to the benefit of the trademark owner, with no intention of infringement of
the trademark. Where such designations appear in this book, they have been printed with
initial caps.
McGraw-Hill eBooks are available at special quantity discounts to use as premiums and
sales promotions, or for use in corporate training programs. For more information, please
contact George Hoare, Special Sales, at george_hoare@mcgraw-hill.com or (212) 9044069.
TERMS OF USE
This is a copyrighted work and The McGraw-Hill Companies, Inc. (“McGraw-Hill”) and
its licensors reserve all rights in and to the work. Use of this work is subject to these terms.
Except as permitted under the Copyright Act of 1976 and the right to store and retrieve one
copy of the work, you may not decompile, disassemble, reverse engineer, reproduce, modify, create derivative works based upon, transmit, distribute, disseminate, sell, publish or
sublicense the work or any part of it without McGraw-Hill’s prior consent. You may use
the work for your own noncommercial and personal use; any other use of the work is strictly prohibited. Your right to use the work may be terminated if you fail to comply with these
terms.
THE WORK IS PROVIDED “AS IS”. McGRAW-HILL AND ITS LICENSORS MAKE
NO GUARANTEES OR WARRANTIES AS TO THE ACCURACY, ADEQUACY OR
COMPLETENESS OF OR RESULTS TO BE OBTAINED FROM USING THE WORK,
INCLUDING ANY INFORMATION THAT CAN BE ACCESSED THROUGH THE
WORK VIA HYPERLINK OR OTHERWISE, AND EXPRESSLY DISCLAIM ANY
WARRANTY, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO
IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. McGraw-Hill and its licensors do not warrant or guarantee that the functions contained in the work will meet your requirements or that its operation will be uninterrupted or error free. Neither McGraw-Hill nor its licensors shall be liable to you or anyone else for any inaccuracy, error or omission, regardless of cause, in the work or for any
damages resulting therefrom. McGraw-Hill has no responsibility for the content of any
information accessed through the work. Under no circumstances shall McGraw-Hill
and/or its licensors be liable for any indirect, incidental, special, punitive, consequential or
similar damages that result from the use of or inability to use the work, even if any of them
has been advised of the possibility of such damages. This limitation of liability shall apply
to any claim or cause whatsoever whether such claim or cause arises in contract, tort or
otherwise.
DOI: 10.1036/0071429042
For my wife, Helena,
and her two brothers and sisters-in-law,
Fernando Errázuriz Guzmán and María Eugenia Oyarzún
José Miguel Errázuriz Guzmán and Mónica Lopez,
who gave me the opportunity to write this book in Chile in 2002
This page intentionally left blank.
For more information about this title, click here.
contents
Introduction to Database Marketing
Acknowledgments
xix
CHAPTER 1
xiii
HOW DATABASE MARKETING WORKS
How It Began
How to Touch a Customer’s Life
How We Have Changed
But Does It Work?
Combining the Database with the Web
Web Response
Customer Service
Computing Lifetime Value
The Retention Rate
Modeling for Churn
Use of Database Marketing for Acquisition
Two Kinds of Databases
Why Databases Fail
Summary
What Works
What Doesn’t Work
Quiz
CHAPTER 2
THE MIRAGE OF CRM
Assumption 1: Why People Buy Things
Assumption 2: Timely and Relevant Offers
Assumption 3: Becoming Customer-Centric
Assumption 4: CRM Mathematics
1
1
3
4
6
6
7
8
9
10
11
12
13
15
17
18
18
18
21
23
24
25
26
vii
Copyright 2003 by Arthur Middleton Hughes. Click Here for Term of Use.
viii
contents
The Loyalty Effect
What to Do with the Data
Summary
What Works
What Doesn’t Work
Quiz
CHAPTER 3
SELLING ON THE WEB
The Web Is a Passive Medium
The Web Is an Ordering Tool, Not a Selling Tool
Web Advertising Was Highly Overrated
Why Web Supermarkets Failed
So, What Is Left? A Huge Research and Transaction Tool
Premium Distribution
Pure-Play Selling on the Web
Brooks: Clever Efforts to Promote Web Sales
Dealer Locator
Citigroup Strikes Out
eToys Strikes Out
Business-to-Business Web Transactions
Selling on the Web: What We Have Learned
Measuring the Value of a Site
What Works
What Doesn’t Work
Quiz
CHAPTER 4
COMPUTING LIFETIME VALUE
Asian Automobiles
Use of Lifetime Value Data
What Works
What Doesn’t Work
Quiz
CHAPTER 5
THE VALUE OF A NAME
The Phone Call
Using LTV to Compute a Name Value
29
32
34
35
35
35
38
41
42
43
43
44
45
48
48
49
50
51
52
53
57
60
60
60
63
64
75
80
81
81
84
85
90
contents
Calculating LTV by Segments
Increasing the Retention Rate
The Value of an Email Name
Your Action Plan
What Works
What Doesn’t Work
Quiz
CHAPTER 6
THE POWER OF COMMUNICATIONS
Business-to-Business Relationship Building
Consumer Relationship Building
Keeping the Advertisers
What Has Happened to RFM?
Do Communications Change Behavior?
Email Communications
Promoting Music Using Emails
Emails from Racing Fans
Stride Rite Shoes
Constructing an Email Test
Summary
What Works
What Doesn’t Work
Quiz
CHAPTER 7
CUSTOMER RESPONSE
“Welcome, Susan”
Setting up a Microsite
How the Web Can Trump the Phone
Generating Web Response
Selling Trucks to Existing Customers
Web-Based Employee Recognition Program
Moving a Loyalty Program to the Web
Summary
What Works
What Doesn’t Work
Quiz
93
93
95
106
107
107
107
109
109
110
111
115
116
118
119
128
130
132
138
139
139
140
142
143
144
146
149
153
155
158
160
161
161
161
ix
x
contents
CHAPTER 8
MARKETING TO CUSTOMER
SEGMENTS
Two Kinds of Data
Reverse Phone Matching
Creating the Segments
Bringing Them Back
Targeting Customer Segments
Quadrant Marketing
Reaching Mothers-to-Be
How Many Segments Should You Create?
Summary of Case Study Results
What Works
What Doesn’t Work
Quiz
CHAPTER 9
HELPING BUSINESS CUSTOMERS
TO BECOME PROFITABLE
Top 10 Prospects
How to Determine the Next Best Product
Computing the 10 Best Prospects
Using the Web Correctly
Postcard Direct
Secret Double Agents
Supply Chain Management
Panduit’s Vendor-Managed Inventory
FedEx Shipping Partnerships
Summary
What Works
What Doesn’t Work
Quiz
CHAPTER 10
CUSTOMER MANAGEMENT
A Little Restraint
Building a Data Warehouse
Risk/Revenue Analysis
164
166
167
170
170
173
175
178
180
183
183
184
184
187
188
190
193
195
200
203
204
208
209
210
211
211
211
214
215
216
218
contents
Loyalty Programs
Automatic Trigger Marketing
Building Loyalty for a Lottery Program
Life Matters
Profitable Use of Customer Data
Dealing with Churn in Cellular Service
Getting Customers to Join a Club
What Works
What Doesn’t Work
Quiz
CHAPTER 11
LETTING THEM COME BEHIND
THE COUNTER
Super Customer Service
Along Comes the Web
Customer Service on the Web
Premier Pages
Letting the Supplier Come behind the Counter
Boeing Spare Parts
A Word of Caution
What Works
What Doesn’t Work
Quiz
CHAPTER 12
CAN DATABASE MARKETING
WORK FOR CATALOGERS?
Use of Database Marketing
Catalogs and the Web
Use of Collaborative Filtering
What Catalogers Have to Guard against When They
Use the Web
Lifetime Value Changes for Catalogers
Building B2B Catalog Sales
Summary
What Works
What Doesn’t Work
Quiz
220
227
231
235
238
239
241
246
246
246
249
250
253
254
256
258
263
266
266
267
267
270
271
272
280
282
282
285
290
290
291
291
xi
xii
contents
CHAPTER 13
FINDING LOYAL CUSTOMERS
Penetration Analysis
Finding Leads for Equipment Financing
Managing Leads on the Web
Building a Relationship with Prospects
Telling Them What They Want to Hear
Cluster Coding
Prospect LTV
Selling Timeshares
Nonprofit Fund Raising
Creating Referrals
Summary
What Works
What Doesn’t Work
Quiz
CHAPTER 14
PROFILE MARKETING
How Modeling Can Help
Getting Customers to Enter Their Own Profiles
The Web Quiz
What Works
What Doesn’t Work
Quiz
CHAPTER 15
A FAREWELL TO THE READER
Database Marketing Works
Summary
APPENDIX
HOW TO KEEP UP WITH DATABASE
MARKETING
ANSWERS TO QUIZZES
Index
357
293
295
296
300
302
305
306
308
310
312
315
317
319
319
320
322
325
328
332
336
336
337
338
338
343
344
346
Introduction to
Database Marketing
D
atabase marketing as we know it today began in the early 1980s.
Its rise was chronicled and boosted by the National Center for
Database Marketing, which was started by Skip Andrew in 1987 and
has continued to hold conferences twice a year ever since. At first, many
companies had heard of it, but few were actually practicing it. By 1995,
however, every American company knew what it was, and most had
appointed a director of database marketing. The practitioners became
like religious converts. They believed in it.
Inside industry, some companies, including Kraft Foods, American
Airlines (and many other airlines), American Express, Hallmark, JC
Penney, Neiman Marcus, MCI, and scores of others, formed huge databases. The idea was that the company could reach its customers directly
(instead of through mass marketing) and build their loyalty by direct
communications. In most cases, it worked. Marketers soon learned how
to set up control groups to prove whether what they were doing was
working. They became skillful at what they were doing.
At first, these large databases were maintained on mainframes, with
smaller databases kept on midrange computers. With the advent of the
PC and servers, however, all large databases moved to servers, and small
databases were maintained on PCs. Companies that kept their data on
mainframes found that they were having considerable trouble keeping
up with the twists and turns of the database marketing revolution.
One twist was the need to create a relational database, which is the
optimum format for database marketing. What that means is that for
xiii
xiv
i n t r o d u c t i o n t o d ata b a s e m a r k e t i n g
each customer, we want to be able to keep all that customer’s transactions and demographics available, organized in tables for easy access and
use in marketing. We append external data and add surveys, profiles,
and preferences. The mainframe programmers just could not keep up.
Today, all large databases are maintained on servers using Oracle or
SQL Server. There are also a number of good database software packages for PCs.
The advent of the Web has made the greatest change in database
marketing since it began. The Web has meant that
• Marketers can access databases directly from their desktop PCs,
using the Web.
• Many companies have Web sites where customers can view and
order products.
• Customers can contact companies not only by calling toll-free
numbers, but also through the Web, which saves companies
millions of dollars.
• Companies can communicate with their customers frequently at
almost no expense by using email. The result is many more
communications, leading to increased loyalty and sales.
• Business-to-business use of the Web has exploded, leading in some
cases to vendor-managed inventory, in which suppliers keep track
of what their customers have in stock and help them to become
successful.
There have been a number of wrong turns, which are described in
this book:
• Selling to consumers on the Web proved to be a major
disappointment, although the Web is an excellent ordering and
customer contact medium.
• Customer relationship marketing (CRM) based on the idea of oneto-one marketing using a massive data warehouse, which many
thought would be an advanced method of database marketing,
proved to be an expensive failure.
• Packaged goods manufacturers discovered that there was no profit,
and in fact considerable losses, from maintaining a database of
ultimate customers. The increase in sales from personal
communication did not pay for the expense of the messages and
the database.
• Marketers learned that they should use their databases to create
customer segments and treat each segment differently. Treating
all customers alike was a loser, since customers differed markedly
in their profitability.
Valuable lessons have been learned. The purpose of this book is to
describe what works and what doesn’t work in database marketing. Too
many books describe only the success stories without talking about the
failures. In each chapter of this book, you will find a list of what works
and what has failed to work.
Figure I-1 What Works and What Doesn’t Work
Marketing to Customer Segments
Lifetime value
Email marketing
Lead development
Profitability analysis
Loyalty programs
Advertising on the Web
Packaged goods DBM
Gold customers
Customer communications
Churn reduction
Web profiling
Penetration analysis
Next best product
CRM data warehouses
Treating all customers alike
xv
xvi
The book contains more than 40 cases from practitioners throughout the United States, Canada, Australia, and Norway. I chose these particular cases because each of them describes the difficulties that had to
be overcome and the way success was measured. Each one makes a
point. Probably the key point is that success does not come from simply doing the following:
•
•
•
•
•
Building a database
Appending data
Buying expensive software
Running models
Communicating with customers
Success comes from developing intelligent strategies that build customer loyalty and repeat sales. To be successful, database marketers have
to think like customers and say, “Why would I want to be in that database? What is in it for me?”
Then they have to dream up strategies that they think will build customer loyalty. They have to test these strategies on a small scale, using
test and control groups. They have to be constantly coming up with
new ideas because they are competing with other marketers who will
copy anything that works and destroy its novelty.
Database marketers constantly work with tiny response rates. Of
course, everyone has heard of marketers who have gotten responses of
20 or 30 percent, but that is very rare. Instead, 2 percent is a good
response rate for a direct-mail offer, and 1 percent is a good response
rate for an email offer. Some marketers get up every day and go to work
knowing that they will not get better than 4⁄10 of 1 percent response on
anything that they do, but they keep on going because in many cases
that rate is enough for profitability.
Incrementalism
One of the first things to remember about database marketing is that
its benefits are incremental. Every company that uses database market-
ing successfully has another profitable sales channel already working for
it, such as retail stores, mass marketing, catalogs, or sales through dealers or agents. The company uses the profits made through these other
channels to build the database and develop the communications that
make it work. The idea in database marketing is that you can increase
your profits by an incremental amount through building customer relationships and information that have the result of
•
•
•
•
Increasing the customer retention rate or repurchase rate
Increasing referrals
Increasing the cross sales and up-sales
Using the customer profile to find more loyal and responsive
prospects
The database does not have to bear the full responsibility for sales—
that is done by the other channels. What the database does is to increase
those sales by a certain percentage. The incremental amount is seldom
massive—perhaps 5 to 10 percent is all. But that 5 percent can be very
profitable.
Marketers have learned how to measure profitability and lifetime
value. There are many examples of this in this book. Marketers have
learned to determine who their Gold customers are and how to retain
them. They have learned to use recency, frequency, monetary (RFM)
analysis to determine which customers are most likely to respond to
offers. They have learned many rules, such as the following:
• Giving customers a choice in a direct offer always reduces
response.
• Don’t market to Gold customers, work to retain them.
• Don’t treat all customers the same way—save your resources to
reward your best ones.
• Always set up control groups to measure your success—if you
don’t, you can’t prove that what you have done has really worked.
• Compute the lifetime value of your customers and put it into the
customer records so that you can use it to evaluate the success of
your marketing strategy.
xvii
xviii
• The more products you sell to customers, the higher their
retention rate will be.
• Customers like to receive communications. Almost any
communication will increase sales and retention rates.
• Referred customers have higher lifetime values and retention rates
than the average customer.
• Offering discounts to attract new customers produces disloyal
customers.
• Use the Web and email as a powerful two-way customer contact
medium.
You will have fun with this book—reading the chapters, taking the
quizzes, and using the charts. All the numerical charts in this book can
be downloaded free from www.dbmarketing.com, where you will find the
answers to the quizzes, more than 100 magazine articles, free software,
and a lot more.
If done properly, database marketing will be fun for both the customer
and the marketer. It is an intriguing game of communication and life.
Enjoy it while you increase your sales and profits.
I will be glad to hear from any readers on any subject at any time. You
may reach me at dbmarkets@aol.com. Point out mistakes in the book,
ask for help with marketing, or send me a case study that I can use in a
future book.
Arthur Middleton Hughes
Vice President for Business Development
CSC Advanced Database Solutions
2100 South Ocean Drive, Suite 16A
Fort Lauderdale, FL 33316
dbmarkets@aol.com
November 2002
ACKNOWLEDGMENTS
I
n addition to the many people mentioned in this book, I want to recognize the special contribution of several individuals who have been
instrumental in providing me with the ideas that have shaped my knowledge of database marketing:
• Paul Wang, Associate Professor of Database Marketing at
Northwestern University. Paul and I gave 28 two-day seminars
together from 1994 to 2000. Paul is an outstanding teacher and a
great friend.
• Bob McKim and Evelyn Schlaphoff at msdbm in Los Angeles,
where I worked for 2 years. Bob and Evie taught me a great deal
about the use of the Web for database marketing. Their ideas
illuminate these pages.
• Fredrick Reichheld, Professor at Harvard Business School, whose
book The Loyalty Effect completely revised my thinking in a
number of crucial areas.
• Beth Clough, Director of Partner Relations at msdbm, a good
friend and an expert on database marketing, who provided valuable
help in the editing of this book.
• Mike Brostoff, CEO of CSC Advanced Database Solutions in
Schaumburg, Illinois, who has run this highly successful database
company for 20 years. Mike and Jeff Lundal of CSC invited me to
work there as vice president for business development. I have been
thrilled to join their team to spread the good word on database
marketing throughout the world.
xix
C
H A P T E R
1
How Database
Marketing Works
T
his is my fifth book on database marketing. My first, The Complete
Database Marketer, was published in 1991. As I write each subsequent book, I am constantly amazed at the changes that have taken
place in the techniques and the uses of database marketing. Today is no
exception. There is so much that is new and different that we are almost
talking about an entirely new marketing method.
How It Began
I like to repeat the story of the genesis of database marketing. It got its
start in a big way in the early 1980s, when mass mailers such as American Express and State Farm Insurance started using their customer lists
to build ongoing relationships with their customers after the initial sale,
leading to increased retention and cross sales. But the roots of database
marketing go back to a period in the United States before there were
supermarkets.
Back in those days, all the groceries in the United States were sold in
small corner grocery stores. The proprietors knew their customers’
names. They would stand at the door and greet their customers by name
as they entered, asking them about their families and their concerns. They
put things aside for customers, helped them carry heavy packages out to
their cars, and built strong and lasting relationships. They built their businesses by developing and cultivating the loyalty of their customers.
1
2
T h e C u s t o m e r L o ya lt y S o l u t i o n
These fellows are all gone today. They were forced out of business by
the supermarkets. Mass marketing took over, along with mass production.
Prices came down. Quality, quantity, and variety went up. The average
corner grocer had 800 stockkeeping units (SKUs) in his store. The average supermarket today has more than 30,000 SKUs. The change affected
the way Americans lived. In 1950 the average American family was spending 31 percent of its household budget on food. Today the average family
is spending only 14 percent on food, and the food it is getting is better
in both quantity and quality than the food it was spending 31 percent on
50 years ago. Because of the lowered cost of food, families have much
more money to spend on hundreds of other products that were out of the
question 50 years ago. So we have all gained.
At the same time, we as suppliers have all lost. We have thousands,
hundreds of thousands, or millions of customers, but we don’t know them,
and they don’t know us. Loyalty has gone out the window. If you talk to
an employee in a supermarket, you are interrupting his regular work,
which is certainly not standing by the entrance and chatting with the customers. You are loyal to that supermarket until tomorrow’s newspaper,
when a certificate from somewhere else leads you to drive to another store.
In the mid-1980s, computers came into wide use. This development
enabled merchants to begin to bring back some of the intimacy that prevailed in the presupermarket days. The software and hardware became
increasingly sophisticated, and their prices have been in a free fall for
years. As a result, it is now possible to keep, economically, in a computer
the kind of information on customers that the old corner grocer used
to keep in his head and to use that to build lasting, profitable relationships with customers.
Many companies have collected huge amounts of data about customers and their transactions, but they have failed to make profitable use of
these data. There is one principle that has remained true throughout the
period since 1985: Database marketing is effective in building customer
loyalty and repeat sales only if the customer benefits from it. The customer has to think, “I’m glad that I’m in that database because . . .,” with
the supplier filling in a meaningful end to the sentence. If the database
does not touch the customer’s life in some way that is satisfying to her,
H o w D ata b a s e M a r k e t i n g W o r k s
she will ignore the communications, leave her gold card behind when
she shops, and refuse to become loyal. Too many companies have
ignored that principle and, as a result, have failed to succeed in database marketing.
How to Touch a Customer’s Life
How do you touch a customer’s life using a database? There are many
ways, some of them so simple that we overlook them. When I fly into
San Francisco, check into a Hyatt hotel, and push the button for room
service, the response is, “Yes, Mr. Hughes. What can I get you to eat?”
I have been in this hotel for less than 20 minutes, yet down in the kitchen
they already know and use my name! This is possible because Hyatt, like
most other hotels today, has Caller ID on its internal hotel telephone
system. When you push that button in your room, the Caller ID goes to
the database that Hyatt created when you checked in and pulls up the
information, “Arthur Hughes, Room 1202” on the screen so that room
service can call you by name. This is database marketing. It is what the
old corner grocers used to do, and it is now made possible by modern
technology used in a creative way. You are 2000 miles from home, yet
people know you and recognize you. It makes you feel great!
Caller ID, of course, is not just for hotels. Customer service departments throughout America are using it to recognize customers when
they make repeat calls. With this technique, before a call is answered,
the database showing the customer’s complete purchase history, demographics, and preferences is brought up on the customer service rep’s
screen, so that he can respond, “Mrs. Webster. So nice to hear from you
again. How did your granddaughter like that sweater you gave her last
October?” This is the kind of thing the old corner grocer used to say,
and we can say it today. But there is a new twist: Using cookies, we are
able to personalize our Web sites so that customers get the same type
of personal greeting when they return to Amazon.com, Barnes &
Noble, Staples, Office Depot, and hundreds of other Web sites. Database marketing today has techniques that make its promise come alive.
3
4
How We Have Changed
There are scores of different examples in the pages of this book. The
principles are the same, but the methods have changed considerably in
the past 15 years:
• When database marketing began, major databases were built on a
mainframe. Today they are built on a server using Oracle or SQL
Server.
• In the beginning, most databases were maintained by the information
technology (IT) department or a service bureau and could not be
accessed directly by marketers. When access was available, it was
by terminals linked to the database by telephone lines. Today,
marketers using PCs access databases over the Web.
• Data used to enter the databases in batch mode as keypunchers
copied customer responses received through the mail. Even telemarketers entered data onto temporary disks, which were later
used to update the mainframe database in batch mode. Today
customers place their orders and update their profiles on the Web,
through a Web page connected with a database on a server. Many
telemarketers also work directly with live data on the same server.
• Database reports used to be produced monthly by programmers
and sent to the users in hard copy. Today, marketers have a menu
of reports on their screens, which they can run every day and print
from their desktop PC printers.
• In the past, to select customers and prospects from a database,
marketers would send detailed memoranda to the programmers.
Upon receipt of these memoranda, programmers would write computer programs to select the desired records and send reports on the
results to the marketers by fax. The process used to take a couple
of weeks. Today, using E.piphany and other advanced applications,
marketers select their own records for promotions using their PCs
through the Web, without the need for IT assistance. The selected
records can be automatically downloaded to the marketer through
the Web or saved for downloading by the mail shop or the e-blasting vendor. The process takes a few minutes.
• Data appending was rare and was done mainly by the largest companies. Today, many companies add demographic data (age, income,
presence of children, credit ratings, and scores of other facts) to
their customer database routinely. Business-to-business marketers
add Dun & Bradstreet (D&B) data (SIC code, annual sales, number
of employees) to all customer and prospect files. These data are used
for segmentation and crafting of marketing communications.
• Customer segmentation was rare. Few companies had identified their
Gold customers, or knew what to do with them once they were identified. Today many companies have segmented their customer base
and have developed different marketing strategies for each segment.
• Communication with customers was mostly by direct mail or
telephone calls. Phone calls cost several dollars each. Letters
were also expensive. As a result, communications with customers
were limited to monthly statements and sales calls. Today,
because of the Web, scores of additional possibilities for communication with customers have opened up. Customers order on
the Web, are thanked for their order, receive a confirming
email, are notified by email when products are shipped, and are
surveyed to determine whether the products arrived and were satisfactory. All these messages were impossible when database
marketing first began. Direct-marketing promotions were sent
out by third-class mail in massive batches. They still are. But to
this has been added email promotions to customers who have
agreed to receive them. These email promotions cost almost
nothing to send, compared to the heav y costs of direct mail.Customer preferences used to be requested through direct-mail or
telephone surveys. These were expensive and, consequently,
rare—they were used mostly for research rather than for customized communication. Today, customers can, and do, complete their personal profiles via the Web, so that outgoing
communications can reflect their preferences.
• Customers used to call toll-free numbers to get information on
product shipments or technical data, and to ask for all sorts of information that was stored in a company’s internal records. They still
do. Increasingly, however, companies are making that information
5
6
available to customers through the Web. This saves the companies
millions of dollars per year and builds customer loyalty by giving
the customers direct access to company archives.
• Models were rarely understood or used. Today all companies with
products that involve churn, such as credit cards and cellular or
long-distance telephone service, routinely use models to predict
and forestall the canceling of service.
This is an exhausting list. It shows where we were and how far we have
come. Perhaps the development that has changed database marketing
most profoundly in the past two decades is the advent of the Web. Today
we can do things that we only dreamed about doing 10 years ago. We can
truly build customer relationships with almost as many personal communications as the old corner grocers used to have with their customers.
But Does It Work?
We can do all these things, but do they really produce more profits? This
is an area in which great strides have been made. Most companies are
acutely aware of the need to use control groups to validate the effectiveness of every single marketing strategy and initiative. In this book you will
find more than 40 case studies that were completed in the past couple
of years by companies that have been able to prove that their techniques
are succeeding in building loyalty, increased sales, retention, repurchase,
and reactivation. Customer communications work. Personalized communications work even better. Now, with the Web, we can afford to
send millions of desired and accepted messages based on data and profiles in our databases that affect customer behavior in a positive way.
Combining the Database with the Web
Perhaps the most interesting shift in database marketing has been the
fact that the database and a Web microsite can be located on the same
server or can be electronically connected so that they seem to be on the
same server. What this means is that the database and the microsite are
not two separate entities, but part of a single customer response and communication vehicle. Cookies are placed on the customer’s computer so
that when she comes to our site, she sees a site that is configured for her,
with things that she has requested to see. This is particularly important
in business-to-business marketing. For example, Dell began the process
by creating Premier Pages for all customers with 400 employees or more.
Dell salespeople contact the purchasing officer at each company and
make a volume pricing deal that is hard to refuse: “We will create a
Premier Page for your company on the Web. If you get your employees
who need Dell products to use this page with passwords controlled by
you, we will give them special prices that are not available to the general public. In addition, we will send you monthly reports on what your
employees are buying and how much you are saving as a result of
your volume pricing deal with Dell.”
Thousands of companies are copying the Dell idea, setting up special
Web pages for their best customers. The system builds sales and loyalty.
Of course each purchase gets an on-screen thank you and confirmation
that can be printed out on the employee’s PC. The employee also gets
a simultaneous confirming email, plus another email when the product
is shipped. She gets a final email survey a few days later asking if the
product has arrived and requesting her opinion of the process. This is
powerful database marketing.
Web Response
Another impact of the Web on database marketing is in the area of
response to communications. Before 1996, customers responded to
marketing communications by phone, mail, or fax. Now more and more,
people are coming to a Web microsite to respond. Because of cookies,
the Web site knows who they are and greets them by name, just as the
old corner grocer used to do.
The microsite is not really a Web site, but part of the database. The
customer enters an order or responds to the survey questions posed in
the communication, and the data go directly into the database. The
7
8
database thanks the customer for the order, which is passed directly to
the Web ordering system. It then sends the customer a confirming email.
A customer who has previously provided credit card information can
order by one click, instead of giving a name, address, and card number
over and over again.
Why do customers prefer this system of response? Because they
never get put on hold. They can get an instant response in 24/7 time
without getting a busy signal or the message “Your call is very important to us. Please stay on the line.” Why do companies prefer this type
of response? Because it saves them millions of dollars. Each call to a live
operator costs a company between $3 and $7 per call. Each response to
a microsite has a variable cost of less than $0.02.
Customer Service
America leads the world in telephone customer service, but we pay a
high price for it. Since the beginning of the database marketing revolution, every company provides a toll-free number for customers to call
with any kind of question. Operators sitting at PCs or terminals linked
to the customer database and to company data and archives receive these
calls through an electronic call director. As customers ask questions, the
operators key in commands on their keyboards and read the answers off
their screens to the waiting customers. They also note in the customer
database that the customer has called, and what the customer called about.
All this expensive service is changing as a result of the Web. Customers are being given access through the Web to the same information that the customer service reps are seeing on their screens. They
are able to enter the same commands and get the same information
without ever being put on hold. They can search for the status of their
shipment, for the price and description of obscure needed parts and
supplies, for the text of previously published articles and white papers.
They can see and print maps, product descriptions, specifications, illustrations, and prices. This service builds customer loyalty and saves the
company millions of dollars.
Computing Lifetime Value
Until 10 years ago, few companies used their databases to compute customer lifetime value. Banks began the process. They were able to determine the profitability of every account on a monthly basis, and to roll that
up to compute household profitability. The results were used to segment
customers in order to learn which ones were useful to the bank and
which ones were reducing the bank’s profits.
I like to use the chart shown in Figure 1-1. It comes from a major bank
in the South that ranked all its customers by profitability. It shows that 80
percent of the bank’s profits came from the top 5 percent of its customers. The bottom 28 percent lost 22 percent of its profits. This is very
powerful information that was just not available before database marketing came along. What was even more powerful was what the bank was able
to do with this information. Clearly the top 5 percent of customers had to
be retained. The bank’s whole future rested on these people. In addition,
something had to change in the bank’s marketing strategy for the bottom
28 percent. Why try to retain people who are costing the bank money?
Not just banks, but insurance companies, business-to-business marketers, and scores of other enterprises began to compute numbers like
these and to develop strategies that made profitable use of them. Many
companies went on to determine potential lifetime value by using
computer models to determine all the other products that each customer
Figure 1-1
Profits by Customer Segment
79.67%
80.00%
60.00%
40.00%
24.82%
15.83%
20.00%
1.52%
0.00%
–20.00%
–21.83%
–40.00%
5%
11%
28%
28%
28%
9
10
was likely to buy from the company in the future. Each such product
had three attributes:
• The potential profitability to the enterprise if the customer were
to buy the product (based on the customer’s assets and income)
• The probability that the customer would buy the product if it
were offered (based on the purchases of other customers with
similar lifestyles)
• The cost to the company of selling the customer the additional
product
From this analysis came the determination of the “next best product”
for each customer (in terms of profit to the enterprise and likelihood of
response). This product was identified in every customer’s database
record and shown on the screens used by all personnel who had customer contact. This has proved to be a highly successful and significant
use of database marketing.
The Retention Rate
The most important number in any lifetime value table is the retention
rate. This is the percentage of newly acquired customers in each segment who will still be buying from your company 1 year later. Before
database marketing, few companies knew what this number was or what
to do about it. I worked with an insurance company whose retention
rate in the first year was only 54 percent. During this year, the company made absolutely no effort to make contact with its customers or
to build a relationship with them. Lifetime value computation showed
that an increase of 5 percent in this first-year retention rate would
increase profits by $14 million (see Table 1-1).
The cost of getting this additional 5 percent increase was estimated
at $36 per customer, or a total of $3 million, for customer communications. The beauty of using a database to do lifetime value analysis is that
you focus on things that you can change by marketing strategy and that
you can prove the value of each strategy in finite dollar terms.
Table 1-1 Effect of 5 Percent Increase in the Retention Rate
New customer
Second year
Second year
Retention rate
Customers
Average revenue
Total revenue
130,000
$2,900
$377,000,000
54%
70,200
$2,900
$203,580,000
59%
76,700
$2,900
$222,430,000
Acquisition/retention
Commission
Total
Total cost
$192
$959
$1,151
$149,630,000
$—
$769
$769
$99,970,000
$36
$769
$805
$104,650,000
Net revenue per customer
Net revenue in total
$1,749
$227,370,000
$1,476
$103,610,000
$1,536
$117,780,000
Using database marketing, the company would not have to commit
the $3 million in the hope of getting this increased return. It could use
the database to set up test and control groups to see if the communications would produce the desired results. It could select 5000 customers
to receive the communications (at a cost of $180,000) to see if their
retention rate could be increased by 5 percent. If the strategy worked,
then the next year the company could apply the successful methods to
all new acquisitions, spend the whole $3 million, and pocket the $14
million. (Table 1-1 and all the other tables in this book can be downloaded free from www.dbmarketing.com.)
Modeling for Churn
As the economy has grown, more companies have learned to use database
marketing to retain customers. Competition in some products, particularly credit cards, cellular phones, long-distance service, and health care,
has become intense. The attrition rate, or the rate at which people switch
to another provider, is very high. At the same time, statistical modelers
have found that regression analysis can help in predicting which customers are most likely to drop the service. The models use a combination
of appended data and transaction history to pick out the variables that precede a customer’s dropping the service. Then advanced database marketing techniques are used to concentrate on these potential defectors and
11
12
Table 1-2 Risk/Revenue Matrix
Probability of leaving soon
LTV
High
Medium
Low
High
Medium
Low
priority A
priority B
priority C
priority B
priority B
priority C
priority C
priority C
priority C
Table 1-3 Price Offers
Number
tested
Number
renewing
Price
offer
Revenue
1000
1000
1000
800
900
950
$10
$ 9
$ 8
$8000
$8100
$7600
get them to change their minds. Defection prediction plus lifetime value
analysis further concentrates the communications on those whom it is most
profitable to serve. The result is risk/revenue analysis (see Table 1-2).
In this analysis, which is discussed later in this book, a concentration
on those customers who are priority A or priority B reduces the task of
lowering churn by 56 percent, saving resources and boosting profits.
Bruce McDoniel of Summit Marketing used this system with a large
regional bank to find the right price for priority A customer renewals.
The goal was to get customers to renew, but also to maximize revenue.
Bruce tried three price offers that netted the bank from $8 to $10. Table
1-3 shows the result of the test.
This simple test enabled the bank to fix the offer price at $9.
Use of Database Marketing for Acquisition
Database marketing is also highly useful for acquiring new customers. The
first step is to apply lifetime value analysis to existing customers to determine the characteristics of the most profitable and the least profitable
customers. This analysis can be used to steer the sales force toward
recruiting the right sort of people in the first place. Frederick Reichheld,
in his path-breaking book The Loyalty Effect, showed us that some people
are inherently loyal, whereas others tend to be disloyal. If companies
concentrate on recruiting loyal customers to begin with, they will have
a much easier time retaining their customers once they are acquired.
In business-to-business customer acquisition, customers can be scored
using Dun & Bradstreet data, such as Standard Industrial Classification
(SIC) codes, annual sales, number of employees, and other such data. By
using this type of appended data, enterprises can learn which types of
companies are most likely to buy which of their products. Their acquisitions can be accurately targeted.
In consumer acquisition, cluster coding has proved useful in many cases.
Cluster coding systems, marketed by Claritas and other providers, segment the U.S. population into 62 different clusters based on lifestyle, with
catchy names like Shotguns & Pickups, Money & Brains, Furs & Station
Wagons, Pools & Patios, and Hard Scrabble. If you apply these clusters to
your existing customer base, you may discover that your services appeal
primarily to 30 percent or so of the 62 possible clusters. Marketing to the
remaining 70 percent may be a losing proposition. By concentrating your
acquisition resources on the prospects that are most likely to buy, you can
increase your success and your profits. Cluster coding works for only a few
products and services, but where it works, it is well worth the money.
Two Kinds of Databases
There are really two different kinds of databases in any company that
is engaged in direct marketing of products and services. One is an operational database, and the other is a marketing database (see Figure 1-2).
An operational database is used to process transactions and get out
the monthly statements.
• For a cataloger, the operational database is used to process the
orders, charge the credit cards, arrange shipment, and handle
returns and credits.
• For a bank, the operational database processes checks and deposits,
maintains balances, and creates the monthly statements.
• For a telephone company, the operational database keeps track of the
telephone calls made and arranges the billing for them.
13
14
Figure 1-2
Operational and Marketing Databases
appended
data
model
scores
RFM &
lifetime value
operational
database
marketing
database
Sales, shipments, payments
Marketing
communications
transactions
general
ledger
promotions
& responses
surveys &
preferences
A marketing database gets data from the operational database, if there
is one. These data consist of a summary of monthly transactions. But the
marketing database also includes much more. It gets additional data from
•
•
•
•
Preferences and profiles provided by the customers
Promotion and response history from marketing campaigns
External sources, such as Experian, Donnelly, and Claritas
Lifetime value and RFM analysis, leading to creation of customer
segments
• Modeling for churn and next best product
The marketing database passes data back to the operational database.
It may advise the operational database
• The segment that each customer has been placed in, which may lead
to operational decisions. Gold customers, for example, may get
different operational treatment.
• Expressed customer preferences, which lead to different
operational treatment; e.g., smoking or nonsmoking rooms are
assigned automatically.
The operational database is run by IT. It is run on accounting principles, and it balances to the penny, since there are legal and tax aspects
to its data. It is audited by external auditors. It contains only current
data on customers. Old data are archived. There are no data on prospects until they make a purchase.
The marketing database is managed by the marketing department.
It is usually outsourced. It is not run on accounting principles, and it
does not need to balance. There are few legal or tax considerations. It
contains information on current customers, prospects, and lapsed customers and communications with them. It may retain data stretching
back over a period of several years.
A data warehouse combines all the data and functions of both an
operational database and a marketing database (plus, in many cases,
other functions such as employee data, payroll, product data, production data, etc.) in one huge database. The data warehouse is so big and
complicated that it needs to be managed by a committee. It is often very
expensive to build and maintain. It is seldom very flexible. Marketing
priorities usually get short shrift in a data warehouse, as other priorities
are often given preference.
Why Databases Fail
There have been many database marketing failures during the past two
decades, from which companies have learned what does not work. A few
of the reasons are
• Failure to test a strategy on a small sample before a rollout. Database
marketing is particularly adapted to setting up tests, which can
prove the validity of any idea before serious money is spent on it.
Database marketers know this. Company management usually does
not. Pushing for quick profits has led many companies to fail with
ideas that could easily have been tested on a small scale first.
• Failure to set aside control groups. If you are going to spend a million
dollars on a group of customer communications designed to boost
retention or sales, you always should set aside a group that does not
15
16
get the communications in order to prove the value of the effort.
Many companies still fail to do this. The result, of course, can be disaster. If the economy is going down and your sales are going down,
you may assume that your methods are not working. If you set aside a
control group, however, you might find that the sales of your test
group went down less than those of the control group, proving that
your communication program worked, even in adverse circumstances.
• Failure to develop strategies that work. Database marketing has often
worked so well that some people have thought that it was the database and the software that were producing the miracles. It wasn’t.
What produced the miracles were strategies that resonated with
the particular audiences represented in the databases. The scarcest
commodities in any database marketing operation are imagination
and leadership: the ability to think up ideas that will work and the
ability to sell those ideas to management and apply them. It is a
highly competitive world today. Friends and Family, created by
MCI 15 years ago, was a wonderfully creative idea for acquiring
new customers. It alone made MCI into a giant long-distance carrier. But once everyone else realized what it was and how it worked,
it was copied by others, and it no longer had the power it once had.
To succeed, you have to constantly come up with new ideas. Some
people just don’t understand this. They think that if they build a
database and apply expensive software, profits will roll in. This is
the reason for the CRM craze that hit the database marketing
world in the previous decade. If you read the CRM manuals, it
almost sounds as if customer acquisition and retention is a simple
matter of building a data warehouse and applying expensive software to it. In general, CRM has been a gigantic failure. We shall
return to this subject in the next chapter.
• Focus on price instead of service. The main reason why customers
stop using a product or service is not the product or the price, but
the way they were treated by the company providing the product
or service. Management always thinks that price is the problem,
hoping that through discounts, it can buy some loyal customers.
Discounts do not buy loyalty. They reduce loyalty. They reduce
margin. They get customers to think about how much they are
paying instead of about how much they are getting. A database
should not be a way to give away coupons. It should be a way to
avoid having to give away coupons because the customers are so
happy with the communications and services they are receiving
that they will ignore the deep discounts of the competition.
• Treating all customers alike. Some customers are more loyal and
profitable than others. Your database should identify these loyal customers, and your marketing strategy should work to retain them
and to encourage others to emulate them. You should also do
something to reduce the number of worthless customers on your
rolls. If you treat all customers alike, you will never improve your
customer mix. A retention budget of $1 million divided by 1 million customers will do very little to change anyone’s behavior.
Divide that $1 million among your 100,000 most valuable customers, and you have $10 per customer per year to reward loyalty and
encourage retention. You may make some progress. Some company
managements do not yet understand this.
Summary
Database marketing is a great success story. During the last 20 years,
thousands of creative and skillful people have developed and applied
new methods for using databases of customers and prospects to acquire
customers, build loyalty, and increase sales. These methods are
explained in detail in this book. But database marketing is more than a
success story. It is a philosophy.
Looked at from the customer’s point of view, database marketing is a
way of making customers happy—of providing them with recognition,
service, friendship, and information, for which, in return, they will reward
you with loyalty, reduction in attrition, and increased sales. Generating
profits by creating genuine customer satisfaction is the goal and hallmark
of satisfactory database marketing. If you are doing things right, your customers will be glad that you have a database and that you have included
them in it. They will appreciate the things that you do for them. If you can
develop and carry out strategies that bring this situation about, you are a
master marketer. You will keep your customers for life, and you will be
happy in your work. You will have made the world a better place to live in.
17
18
What Works
•
•
•
•
•
•
•
•
•
Getting customers to respond to you on the Web
Creating premier pages for your best business customers
Providing customer service information free on the Web
Determining the lifetime value of your customers and using it to
evaluate your marketing strategy
Determining the next best product for each customer and putting
it in the database for your employees to see and use
Learning the retention rate for each of your customer segments
and working to improve it
Sending email communications to customers who have asked for
them
Using cookies to recognize and greet customers when they come
to your Web site
Using risk/revenue analysis to concentrate your marketing
resources on customers whose behavior you need to modify
What Doesn’t Work
• Building a huge data warehouse for the purpose of building
profitable relationships with your customers
• Treating all customers alike
• Failing to set up control groups
• Failing to test any new idea on a small group first
• Failing to use the Web as part of your database communications
program
Quiz
1. Which of the following has been a failure in database marketing
since 1985?
a. RFM
b. The Web
2.
3.
4.
5.
6.
c. Treating all customers alike
d. Email
e. Direct mail
In the case given in this chapter, it was suggested that the
increased retention was due to
a. Customer communications
b. Advanced software
c. CRM
d. More loyal customers
e. Cookies on the Web
What is considered to be a good average response rate for email
communications?
a. 30 percent
b. 20 percent
c. 2 percent
d. 1 percent
e. 0.4 percent
What is the most important factor in database marketing success?
a. Software
b. A data warehouse
c. The Web
d. Marketing strategy
e. CRM
In risk/revenue analysis, the marketer concentrates most on
a. Those customers in priority A and priority B
b. The lifetime value of priority B customers
c. The churn model
d. The lifetime value model
e. The data warehouse
What’s wrong with treating all customers alike?
a. It’s undemocratic.
b. The resources available to devote to the best customers
will be too small.
c. Even worthless customers deserve a break.
d. It is impossible to segment customers by behavior.
e. Customers would not like it.
19
20
7. Where is a microsite for customer response ideally located?
a. On the main company Web site
b. On a PC
c. On a mainframe
d. On the computer holding the database
e. In a risk/revenue matrix
8. Which is not true of the MCI Friends and Family program?
a. It worked very well for a few years.
b. It was copied by other companies.
c. It made innovative use of the Web.
d. It was a cost-effective way to add customers.
e. It helped MCI to become a telecom giant.
9. What is a database marketing control group?
a. Customers who get a special test group of
communications
b. Those who dictate database marketing strategy
c. Top management
d. Those who do not get any of the test communications
e. A privileged customer group
10. Discount certificates are
a. Best distributed by direct mail
b. Used for building customer loyalty
c. Used to boost margins
d. Used to ward off competition
e. Not a good way to recruit new customers who last
11. What one development has changed database marketing
profoundly in the past two decades?”
a. CRM software
b. Data warehouses
c. Desktop PCs
d. the Web
e. Customer lists
C
H A P T E R
2
The Mirage of CRM
Before implementing a CRM package it is necessary to understand why
nearly half of U. S. implementations and more than 80% of European
implementations are considered failures. It’s difficult to fathom failures
of such monstrous proportions, especially when complete CRM installations can cost millions of dollars and then hundreds of thousands of
dollars more per annum.
John_taschek@ziffdavis.com
C
ustomer relationship marketing (CRM) burst on the marketing
world in the early 1990s. It was soon adopted by some of America’s
largest corporations. After working with it for several years, most analysts admit that it does not seem to deliver the benefits that it promised
and that it may, after all, be a mirage. Let’s see what it was supposed to
do and why it failed.
The basic idea was to collect a great deal of information about customers, prospects, products, promotions, and other such things and put
it into a giant in-house data warehouse. Many of these warehouses were
very expensive, costing as much as $20 million. Part of the cost was that
of appending external data about customers and prospects, such as age,
income, presence of children, lifestyle, home value, creditworthiness,
and so on. To access and manipulate the data in the warehouses, very
sophisticated software, costing at least a million dollars the first year,
21
22
was installed. In many companies there was a tendency to focus on
information technology (IT) rather than on the business outcome.
CRM was introduced by the IT unit, often without sufficiently close
coordination with marketing, sales, or customer service. The result was
expensive systems that did not deliver either successful campaigns or
improved profits.
The National Retail Federation’s annual CRM Conference survey of
North American retail companies found that 69 percent of the respondents said that they had gained little or no benefit from their CRM
investment. Why was this so?
What was CRM designed to do? There were really two basic goals:
• Build and maintain relationships with prospects and customers
based on having lots of relevant information about them, and using
that information to guide communications and contacts with the
customer.
• Use that information to make the right offer to the right customer
at the right time, thereby increasing sales and pleasing the
customers.
The idea that these goals could be achieved was based on four
assumptions, all of which proved to be invalid in practice:
1. One-to-one marketing is an achievable goal given the right
information and software. Stated another way, customers’ and
prospects’ purchasing behavior can be predicted accurately from
the information that you can collect about them.
2. Customers’ and prospects’ behavior is heavily influenced by
timely and relevant offers.
3. Companies can shift rapidly from being product-focused (with an
employee’s bonus being based on the number of particular products
sold) to being customer-focused (with an employee’s bonus being
based on the amount sold to particular customer segments,
regardless of what products they buy).
4. Introduction of CRM has a positive return on investment because
it increases sales and profits by more than its cost.
T h e M i r a g e of C R M
Assumption 1: Why People Buy Things
Let’s see why none of these assumptions proved to be correct. In the
first place, why do customers and prospects decide to buy something?
You can come up with lots of answers, but the key answer is that they
decide that owning the product or service at a particular time will make
them happier than not owning it. What happiness consists of is defined
by the person who is trying to possess it, and it varies from person to
person and from time to time.
I like Big Macs, but not all the time. Sometimes I like fish, sometimes
I like chicken, and quite often I like home-cooked vegetables. There is
no way that you can accurately predict what I will want to eat next unless
you know what I ate last, how hungry I am, where I am at the time, whom
I am with, and how much money and time I have available right now for
eating. There is no way that any data warehouse could ever collect such
timely and relevant information or that anyone other than the customer
could accurately weigh the importance of each piece of information.
The same principle applies to predicting my interest in taking a trip,
buying clothing, buying a car or an appliance, or taking a college course.
You can collect some relevant information, but you cannot collect
enough to make accurate predictions concerning what Arthur Hughes
will do today. What you can say with some accuracy is that people in
Arthur Hughes’s age and income group who have similar purchasing
habits are more likely to buy a certain type of clothing or insurance policy than the average person picked out at random. But of course, that
is not CRM or one-to-one marketing. That is regular database marketing, where you are targeting segments. You don’t need a data warehouse for that. You need only a database (which can be built for a
fraction of the cost of a data warehouse) that permits dividing the customer and prospect base into purchasing segments.
What is the difference between database marketing and CRM?
They are both based on databases of prospects and customers
that are used to guide marketing and sales strategy. CRM requires
23
24
a large data warehouse with costly software, aimed at determining
and influencing the behavior of individuals through one-to-one
marketing. Database marketing is based on a data mar t, which
costs about 10 percent of the amount required for a CRM
warehouse. Database marketing is usually aimed at identifying
customer segments and marketing to them, and also building
one-to-one relationships with existing customers through loyaltybuilding communications.
To see the difference, think of 100,000 customers on whom you
have data. With database marketing, it might be possible to predict
with 75 percent cer tainty that 40 percent of a cer tain segment of
those customers will buy product X within the next month. But
you cannot be 75 percent cer tain that Ar thur Hughes (a member
of the segment) will buy the product next month—or ever, for
that matter. The first conclusion comes from segment analysis
using database marketing. The second conclusion would come
from CRM if it could be made to work. Unfor tunately, it cannot.
We should note here that there has been a corruption in terminology.
Many companies are practicing database marketing, but are calling it
CRM because that sounds more modern and up to date. They don’t have
a data warehouse. They don’t have million-dollar software. They don’t
really even attempt one-to-one marketing. They have just built a modest data mart and are creating marketing segments and marketing to
them. They call what they are doing CRM, but it is really plain old database marketing, which works wonderfully in the right situation.
Assumption 2: Timely and Relevant Offers
The second assumption is wrong for a similar reason. Do you eat
because you are hungry, or because you received a timely communication telling you that a particular restaurant was having a special on lobster dinners? Of course, if you are hungry, you enjoy lobster, and you
have the time and money to go to that restaurant right now, the offer
would be a delightful blessing. But the warehouse cannot contain that
information. Mass marketing on TV showing people eating delicious
lobster dinners at a low price can work wonderfully. The image sticks
in people’s minds and will be recalled the next time they are hungry.
You don’t need a warehouse to design and place such ads. Mass marketing works very well and is a lot cheaper in terms of sales per dollar spent.
Of course, there are situations in which a timely and relevant offer
can make the difference in a sale. Banks know a lot about people’s
demographics and behavior. They can predict the probability of a particular customer’s buying a particular product and the lifetime value
that the customer would produce for the bank if it were to offer the
product and the customer were to accept the offer. This also applies
to automobile companies that have customers on 3-year leases, and to
scores of other situations. But for most of these situations, good old
database marketing will work just fine. We can pick out all the people
who have lived in their homes for 5 years or more and do not have a
home equity loan, and send them an offer. It works. We are marketing
to segments. We don’t need a warehouse or one-to-one marketing to
be successful.
Assumption 3: Becoming Customer-Centric
The third assumption, that companies can rapidly move from being
product-focused to being customer-focused, is the weakest of all. After
working with dozens of large American corporations, I can report that
it is almost impossible for most companies to become customer-focused.
Why is this so? Because of the way business is developed. You create a
new camera, a new computer, a new automobile, or a new hotel. The
company that develops the product wants to realize a return on its
investment. It puts a product manager in charge of marketing the product and tells her to get busy.
The product manager does everything she can think of: mass marketing, point-of-purchase displays, retailer subsidies, direct marketing.
Of course, if you have a list of prospects in your warehouse, she may
try a promotion to that list to see if it works. But her goal is to sell hotel
space in this new hotel, and only secondarily to increase overall customer lifetime value to the company.
25
26
Could you replace all the product managers with customer segment
managers and have no product managers at all? Not likely. No company has ever done that, and none ever will. Once you have developed
a product, you can’t leave the promotion of that product to chance. You
simply won’t say, “We will be nice to our customers and hope that they
notice our new hotel. We won’t specifically promote this new hotel,
because one of our older hotels may suit some customers better.”
Assumption 4: CRM Mathematics
To understand the return on investment from CRM, let’s do a little
math. We have to start with some basic principles. Customers do
respond to communications, offers, and promotions. For the communication to work, however, the seller has to have some sort of established
business. He has a product or service to sell and a working distribution
channel. In order to use CRM, he has to have a revenue stream (otherwise, where would he get the money to build his warehouse?).
The reason for using CRM is to make incremental increases in the level
of sales that is currently being made through mass marketing, retail
stores, catalogs, and database marketing. The marketer assumes that if
there were just more information about the customer’s decision-making
process, it would be possible to increase the sales level (of customers and
prospects) by as much as 5 or 10 percent. Let’s say that the marketer is
selling some product or service for $150. It could be rental cars, small
appliances, or carpets. The marketer contacts 20 million customers per
year and has sales of $900 million. The current marketing methods
involve mass marketing and direct marketing to customer segments,
including a loyalty program. We will assume that of the $150, net profits per sale are 10 percent after deducting all costs, including marketing.
Pricing the CRM Program
We will assume a modest CRM warehouse containing data on 14 million prospects and past customers and 6 million current customers. Let
us assume that we can build it for $10 million, including the software
for access, with annual maintenance costs of $1.5 million, which include
the costs for the staff, NCOA (National Change of Address), and
appended external data. Amortized over 3 years, the warehouse will cost
us $5.5 million per year, as shown in Table 2-1.
Let’s compare this to the cost of database marketing in the same situation, shown in Table 2-2. It costs far less. We are assuming a cost of
$1.5 million to build the database. Many databases are built for far less,
such as $500,000, and some, of course, cost more. The number $1.5
million is a good number for a large corporate marketing database.
Cost of Communications
To these CRM or database costs, we must add the cost of communications
with customers and prospects. After all, if we are going to make the
right offer to the right customer at the right time, we have to communicate the offer to customers personally, or what is the warehouse for?
Let’s assume that our firm has been collecting email names to keep the
cost of communications down. It has email names for half of its customers. Table 2-3 shows the total CRM and database cost, assuming an
average of only three communications per customer per year. Note that
when CRM or database marketing is used properly, some likely customers may get more than four communications a year, and some will
get only one.
Table 2-1 Warehouse Cost
Cost
Cost to build warehouse
Amortize over 3 years
Annual maintenance
Annual cost
$10,000,000
$ 4,000,000
$ 1,500,000
$ 5,500,000
Table 2-2 Database Cost
Cost
Cost to build database
Annual maintenance
Annual cost
$1,500,000
$ 600,000
$1,000,000
$1,600,000
27
28
Table 2-3 Comparison
Communications
Quantity
Rate
DB/warehouse
Mail communications
Email communications
CRM/database costs
3,000,000
3,000,000
$0.60
$0.04
Per
year
CRM
annual cost
Database
annual cost
2
4
$5,500,000
$3,600,000
$ 480,000
$9,580,000
$1,600,000
$3,600,000
$ 480,000
$5,680,000
Table 2-4 Net Profit
Customers
Annual sales per customer
Sales
Profit per sale
Gross profit
CRM/database cost
Net profit
Without CRM
or database
marketing
With CRM
With
database
marketing
6,000,000
$150
$900,000,000
10%
$90,000,000
$0
$90,000,000
6,000,000
$165
$990,000,000
10%
$99,000,000
$ 9,580,000
$89,420,000
6,000,000
$165
$990,000,000
10%
$99,000,000
$ 5,680,000
$93,320,000
So with these costs in hand, what will be the result of using CRM or
database marketing? See Table 2-4.
Using either CRM or database marketing, we have managed to
increase the average sales per customer by 10 percent, to $165. We may
have done this by getting some customers to place larger orders, by getting some to place more orders, by reducing the attrition rate, or by a
combination of these and other methods. Sales have increased by $90
million, with increased gross profits of $9 million. Unfortunately, when
we factor in the cost of CRM, net profits have gone down by $580,000.
That is what virtually all companies that installed CRM have discovered. Note that these figures assume that CRM is working—that it is
in fact increasing sales by 10 percent. But in many cases, for the reasons given earlier, CRM has no such valuable benefit. In the majority
of companies, it does not affect the sales rate at all. Much of the money
spent on CRM is wasted. That is the reason for the Gartner Group’s
pessimistic conclusion on CRM.
The methods whereby database marketing produces the gains in
profit shown in this section are explained in the remaining chapters and
case studies in this book. A further discussion of the benefits of a data
warehouse is given in Chapter 10, “Customer Management.”
The Loyalty Effect
CRM is often justified by citing references to reports like those of Bain
& Company, which maintain that companies can boost customer net
present value by as much as 95 percent if they can retain only 5 percent
more of their customers. Numbers like these have often been used to
persuade management that a CRM package could pay for itself very
rapidly. The assertion that growth in net present value results from
increased retention is central to Frederick Reichheld’s excellent book
The Loyalty Effect. Figure 2-1 shows Reichheld’s increases in customer
net present value in various industries as a result of a 5 percent increase
in the retention rate.
These numbers seem too good to be true. In fact, they are too good
to be true. Let’s see what the real numbers might actually be. Let’s take
one industry from the list as representative of the problem: life insurance. We will take a typical term life insurance policy of $500,000 for
a 45-year-old nonsmoker that costs $1000 per year. For such policies,
Figure 2-1
Reichheld 5 Percent Retention Gains
Advertising Agency
Life Insurance
Publishing
Branch Bank Deposits
Auto/Home Insurance
Auto Service
Credit Card
Industrial Brokerage
Industrial Laundry
Industrial Distribution
Office Building Management
Software
95%
90%
85%
85%
84%
81%
75%
50%
45%
45%
40%
35%
0%
20%
Impact of a 5-percentagepoint increase in retention
rate on customer net present
value
40%
60%
80%
100%
Source: Frederick Reichheld, The Loyalty Effect (Boston: Harvard Business School Press, 1996), p. 36.
29
30
Table 2-5 Life Insurance Net Present Value
Year 1
Year 2
Year 3
Year 4
Year 5
Retention rate
Customers
Payments
Revenue
50%
100,000
$1,000
$100,000,000
60%
50,000
$1,000
$50,000,000
65%
30,000
$1,000
$30,000,000
70%
19,500
$1,000
$19,500,000
75%
13,650
$1,000
$13,650,000
Commissions
Claims
Total costs
$ 20,000,000
$ 75,000,000
$ 95,000,000
$ 250,000 $ 150,000
$37,500,000 $22,500,000
$37,750,000 $22,650,000
$
97,500
$14,625,000
$14,722,500
$
68,250
$10,237,500
$10,305,750
Profit
Discount rate
Net present
value
Cumulative
NPV
$
5,000,000
1.00
$12,250,000
1.07
$ 7,350,000
1.14
$ 4,777,500
1.21
$ 3,344,250
1.29
$
5,000,000
$11,491,557
$ 6,468,043
$ 3,943,929
$ 2,589,822
$
5,000,000
$16,491,557
$22,959,601
$26,903,530
$29,493,352
the first-year retention rates are very low. Subsequent rates are much
higher. Table 2-5 shows the net present value for such policyholders for
the first 5 years.
This chart shows that the cumulative net present value (the present
value of the profits) from 100,000 new policyholders will be $29.5 million
over 5 years. Note that we are assuming that 50 percent of the policyholders drop their policies during the first year. After that year, the
retention rate increases. Of particular importance is the discount rate.
This is the rate used to discount future profits and convert them into
net present value profits. The formula for the discount rate is
D ⫽ (1 ⫹ i * rf ) years to wait
In the third year (2 years to wait), for example, the formula is
D ⫽ (1 ⫹ 0.06 * 1.1)2
This becomes
D ⫽ 1.14
In this formula, i is the market rate of interest, figured at 6 percent,
and rf is the risk factor, which is determined to be 1.2 for the first year and
1.1 for subsequent years. The risk factor is a number that represents the
risk that you won’t generate the sales and profits at all. A perfectly safe
business has a risk factor of 1. The risk can come about because customers don’t pay, because of heavy competition, because of product obsolescence, etc. Years to wait is the amount of time that the insurance company
has to wait before the profits materialize. The first year, we are assuming
that the waiting time is zero, because we are asking the policyholders to
pay in advance. The chart could be made more complicated by adding
administrative costs, medical examinations, etc. Leaving these out, we
have a relatively clear picture of the marginal profits that a life insurance
company can expect to realize over 5 years from 100,000 policyholders.
Using the Bain & Company figures, we should expect to see this
$29.5 million 5-year profit go up by 90 percent, to $56 million, if we
were to increase the retention rate by 5 percent. But that is not what
happens. Table 2-6 gives the profits over 5 years assuming that the
retention rate is somehow increased by 5 percent.
The cumulative net present value does go up, but only by about 15
percent, as Table 2-7 shows.
Now, 15 percent is a good solid number, but it is not 90 percent.
Unfortunately, the rate is probably much less than 15 percent, because
we have not figured in the costs of getting the 5 percent increase in
retention. Here is where CRM really has trouble.
Table 2-6 New NPV
Year 2
Year 3
Year 4
Year 5
Retention rate
55%
Customers
100,000
Payments
$1,000
Revenue
$100,000,000
Year 1
65%
55,000
$1,000
$55,000,000
70%
35,750
$1,000
$35,750,000
75%
25,025
$1,000
$25,025,000
80%
18,769
$1,000
$18,768,750
Commissions
Claims
Total costs
$20,000,000
$75,000,000
$95,000,000
$ 275,000
$41,250,000
$41,525,000
$ 178,750
$26,812,500
$26,991,250
$ 125,125
$18,768,750
$18,893,875
$
93,844
$14,076,563
$14,170,406
Profit
Discount rate
Net present
value
Cumulative
NPV
$5,000,000
1.00
$13,475,000
1.07
$ 8,758,750
1.14
$ 6,131,125
1.21
$ 4,598,344
1.29
$5,000,000
$12,640,713
$ 7,707,752
$ 5,061,375
$ 3,561,005
$5,000,000
$17,640,713
$25,348,465
$30,409,840
$33,970,846
31
32
Table 2-7 Gains from 5
Percent Increase
Present 5-year NPV
New 5-year NPV
Increase
Percent increase
$29,493,352
$33,970,846
$ 4,477,494
15%
How does Reichheld suggest that the 5 percent increase be brought
about? He has some very ingenious and valid ideas. The book is filled
with excellent suggestions. Here are only a few of them:
• Improve employee loyalty by increasing the employee retention
rate (since retained employees help to retain customers).
• Recruit the right sort of customers to begin with (some customers
are inherently loyal, and others are inherently disloyal).
• Change the commission system to reward retention as much as or
more than acquisition.
Nowhere in Reichheld’s book does he suggest building a huge data
warehouse and purchasing CRM software to go with it. If we had done
this, we would have had to factor into Tables 2-6 and 2-7 the cost of the
CRM needed to produce the 5 percent increase in the retention rate.
The 15 percent growth in net present value might disappear or even
become negative if we were to do so.
So, unfortunately, CRM proponents have been using numbers from
Reichheld’s book to justify their programs without any solid justification
in the book whatsoever.
What to Do with the Data
There is a fundamental disconnection between the goals of CRM and the
results. If you have invested millions of dollars and you have a wonderful
data warehouse and analytic software, then what? How do the data warehouse and software increase the retention rate? Somehow, you are supposed to take the software and change customers’ behavior. The CRM
Forum (www.crm-forum.com) lists three reasons why CRM usually fails:
1. Problems with organizational change and internal politics
2. Lack of the right skills and enterprisewide understanding of the
initiatives
3. Poor initiative planning
Unfortunately, many CRM analysts write a type of dense prose that is
almost impossible for normal people to understand. Take this example
from a Gartner Group report on the CRM Forum:
Enterprises struggling with lack of coordination across departments
should consider the needs of individual groups and work on a topdown/bottom-up strategy. Because department-level CRM deployments
may be more cost-effectively supported by a Tier 2 system integrator
with domain-specific skill sets and software relationships, the enterprise
should identify a master list of approved smaller service providers to
work on individual projects.
It really should not be so complicated. The goal of CRM is to select
the right prospects and to get customers to be more loyal and buy more.
Software is of only marginal help in doing this. The goals can be
achieved only by coming up with products, services, tactics, strategies,
and communications that the customers will like and respond to. The
CRM warehouse and software do none of these things. The goal can be
realized, however, in many other ways, which are outlined in this book.
So, what should you do? Sungmi Chung and Mike Sherman suggested in the McKinsey Quarterly that the two most essential resources
in any CRM program are a wealth of hypotheses about the composition of possible target segments and a variety of offers to test these
hypotheses. They point out that
Less is more if it makes it possible for the process of testing and refining
offers to start sooner rather than later. It will be more than enough to
begin with some demographic details about customers and information
about the products they currently own, such as their account balances,
and the kinds of transactions they have undertaken during the course
of the previous 12 months.
33
34
Summary
CRM is, in general, a mirage, an illusion based on a nice idea (the right
offer to the right customer at the right time) that does not work out in
practice:
• One-to-one marketing, which sounds wonderful, may in fact not
be possible in most product and service situations. We lack the
information. The data about any individual that are available may
not be sufficient to predict that individual’s behavior, although it
may be sufficiently accurate to predict the behavior of the
customer segment of which he or she is a part.
• Making the right offer to the right customer at the right time may
be impossible. People make decisions based on personal factors
that cannot be captured in a database. We can, however, profitably
target segments.
• Companies cannot easily shift from being product-focused to
being customer-focused. It sounds great, but few have actually
done it. Most never will.
• Even if CRM were to work in theory, it does not work in
practice because the increased profits are swamped by the
increased costs.
• We must keep in mind that customer communications, which are
triggered by CRM or a database, produce only incremental
results. You are already selling something. You hope to sell more
and keep customers longer by using the database and
communications. The incremental sales are usually only slightly
more than the sales you would have gotten without the database
and the communications. You have to justify your CRM not on
the basis of total sales, but on the basis of these incremental sales.
• Many marketers are doing database marketing and calling it CRM.
They don’t have a warehouse or million-dollar software. They are
marketing to segments and making money doing it. That is great.
But it is not CRM.
• NCR, which sells the TeraData computer, has developed some
very sucessful data warehouse applications for very large banks,
insurance companies, and others; two of these are described in
detail in this book. These applications are an exception to the
general rule about data warehouses. NCR has succeeded where
most other CRM developers have failed.
What Works
• Mass marketing for certain products and services
• Building a customer and prospect database and using it to market
to customer segments for certain products and services
• Having your products available in retail locations that are
convenient for your customers
• Communicating with customers to maintain their loyalty; sending
them emails, thanking them for their patronage, and getting them
to fill out customer preference profiles
What Doesn’t Work
• Giant CRM multimillion-dollar data warehouses designed to
produce one-to-one marketing.
• Million-dollar software that is designed to make the right offer to
the right customer at the right time.
• One-to-one marketing, except in certain limited situations.
• Changing a company so that it is customer-focused instead of
product-focused. It’s a nice idea, but no one has ever done it, and
no one ever will.
Quiz
1. Which is not a basic assumption of CRM?
a. A warehouse permits a firm to make the right offer to the
right customer at the right time.
b. A data warehouse is essential to marketing success.
35
36
2.
3.
4.
5.
c. One-to-one marketing is achievable.
d. Companies can become customer-focused.
e. Increased employee loyalty is essential to increased
customer retention.
The formula for the discount rate includes all but which of the
following:
a. The waiting time
b. The market rate of interest
c. The marketing costs
d. The risk factor
e. The years to wait
Which of the following is not given in the chapter as a reason
why CRM is a failure?
a. CRM software is not available today.
b. The data to predict customers purchasing behavior are
lacking.
c. Data warehouses are very expensive.
d. Companies cannot become customer-focused.
e. The Loyalty Effect does not support CRM.
Turn the following paragraph into English:
Enterprises struggling with lack of coordination across departments should consider the needs of individual groups and work on
a top-down/bottom-up strategy. Because department-level CRM
deployments may be more cost-effectively supported by a Tier 2
system integrator with domain-specific skill sets and software relationships, the enterprise should identify a master list of approved
smaller service providers to work on individual projects.
What does the National Retail Federation’s CRM Conference
survey of North American retail companies estimate the CRM
failure rate to be? Nearly
a. 80 percent
b. 70 percent
c. 60 percent
d. 50 percent
e. 40 percent
6. Which of the following was not listed in the chapter as a
recommendation by Frederick Reichheld in The Loyalty Effect?
a. Improve employee loyalty
b. Build a data warehouse
c. Recruit the right sort of customers to begin with
d. Change the commission system
e. Increase the customer retention rate
7. What does NPV stand for?
a. Not previously viewed
b. Net present value
c. The rate of profits from an enterprise
d. The cumulative profits expected in the future
e. The discount rate
8. What does it mean to say that database marketing and CRM
produce only incremental results?
a. They don’t work.
b. They don’t work very well.
c. The basic product sales are produced by other channels.
d. A 5 percent increase in the retention rate produces a 95
percent increase in profits.
e. None of the above.
9. Why will most companies never become customer-focused?
a. Because companies are designed for customer acquisition
b. Because Reichheld said it cannot work
c. Because companies want to see that individual products
succeed
d. Because the compensation system cannot be changed
e. Because of test and control groups
10. What does the book suggest as the best way to attract customers
to Red Lobster?
a. Database marketing
b. CRM
c. Mass marketing
d. Certificates
e. None of the above
37
C
H A P T E R
3
Selling on the Web
The number of online 13- to 22-year-olds will grow from 17 million
to 22 million by 2004. How can your company target a group that,
unlike adults, internalizes the Internet? What do you need to do to
attract and keep them as customers? How should you alter your marketing and product development strategies? Forrester’s Wired Youth
Summit will help you understand this sought-after group and give you
new ideas you can use today to develop these critical relationships.
W
hen the Web arrived in the mid-1990s, just about everyone
in marketing thought it would be a great sales medium to
consumers—something like TV, radio, or direct mail. Billions and billions of dollars were invested with that idea in mind. Thousands of new
companies sprang up to sell on the Web or to provide services to those
who would be selling there. Every major corporation in America produced a Web site, most of which involved the marketing of some product
or service. Hundreds of sites planned to make money from advertising
placed on their sites by others. Spending on the Web was a big part of
the boom times of the nineties.
In fact, between 1995 and 2002, very few companies were selling
very much to consumers on the Web. With the exception of travel,
music, pornography, and books, the sales were very, very small. Why
were they so small? At first, people said it was because half the nation
was still not on the Web (hence millions of dollars were spent to equip
38
Selling on the Web
schools for the Web). Then the argument ran that even those who were
on the Web were hesitant to buy because of the possibility of credit card
fraud. No one was saying that the Web was not a great sales medium.
Everyone just said, “These things take time.” Amazon.com, the leader that
showed everyone how Web sales should be done, didn’t make a profit
until 2001. But Amazon’s failure was said to be due to its overextension
into too many products. If it had just stuck to books, it could have been
profitable, everyone agreed.
The quote at the beginning of this chapter is really a joke. Here is a
“wired youth summit” that teaches you how to sell to young people who
have “internalized” the Web. I could be wrong, but my experience tells
me that in most cases these consumer sales will not materialize for the
reasons given in this chapter. Teaching retailers how to reach these “wired
youths” through the Web is a futile exercise. Amazon was the first Web
bookstore. Within 3 years of its debut, almost everyone in the middle class
in America had heard of it and had used it. It was impossible for any
“pure-play” Web bookstore to start up and succeed in competition with
Amazon. Established booksellers such as Barnes & Noble and Borders
opened up competing Web sites. In other fields, retailers rushed to
become the first mover. More than a dozen online supermarkets were
started, such as Peapod, Web Van, and Web Grocer. Many travel sites
opened up: Travelocity, Expedia, Priceline. Banks, insurance companies,
discount brokers, department stores—every type of business rushed to
be the first and the best on the Web.
As people began to see that the sales just weren’t materializing, they
began to try to understand why. There was the debate between the “pureplay” and the “bricks and clicks” merchants. Pure-play referred to Web
sellers who had no retail stores. At first, Amazon was pure-play, as were
E*TRADE and Priceline.com. Some said that success at Web selling
would more likely come to established chains that had bricks-and-mortar
enterprises, could offer local delivery, were used to filling orders in quantities of one, and understood retailing. Barnes & Noble, by this reasoning,
should be much more successful than Amazon because Barnes & Noble
had thousands of retail stores. But, despite the debate, only AOL, eBay,
and a few others, whether pure-play or bricks and clicks, were making a
profit. Everyone knew that it would all work out some day, but when?
39
40
Part of what was pushing everyone to spend like mad on the Web was
the idea of the “first mover advantage.” America Online gave away six
diskettes to every man woman and child in America and soon not only
became number one, but absorbed most of its rivals and then took over
the giant Time Warner conglomerate. AOL was always profitable. But
several unprofitable dot coms had successful IPOs, making their backers
very wealthy. After that, investors went absolutely crazy over the Web.
Many of these Web companies focused on the “new economics.”
Profits were unimportant to the new economics. The important things
were hits and clicks. How many people visit your Web site? How long
do they stay there? What do they look at? What do they click on?
Those who pointed out that the Web sites did not show any profits were
considered “old-fashioned” people who “just didn’t get it.” Many of the
Web firms were run by young people, many of them technically
inclined, who had never run a business of any kind before. They wore
jeans and ponytails and shopped for organic foods. It was a weird, wonderful world, and it lasted about 5 years.
It all came to a halt in May 2000 when the judge’s decision in the
Microsoft case came down. While Microsoft was the biggest software
company in the world, it was not a dominant presence on the Web. But
Microsoft, in everyone’s mind, was at the heart of the boom of the
1990s. Most people felt that the boom was due to the PC revolution,
which Microsoft was given credit for starting and servicing. The Web
was the second step in the PC revolution. If the government was going
to break up Microsoft, what was next? Perhaps the whole boom was on
shaky ground. Venture capitalists began to wonder if they would ever
get back the hundreds of billions of dollars that they had invested in
Web companies—99 percent of which had yet to show a return, or a
hope of a return.
That was when hundreds, then thousands, then millions of people
began to realize and admit that the emperor had no clothes. Investors began
to look more closely at their Web investments, and they did not like
what they saw. Microsoft, after all, was one of the most profitable companies in the history of the world. But only a handful of the hundreds
of thousands of Web sites were profitable at all. And, worse, few of them
had any hope of profitability in the near future. Most of them even
Selling on the Web
lacked a long-term profit forecast. Selling on the Web to consumers
turned out to be a gigantic failure. Let’s see why that was.
The Web Is a Passive Medium
You listen to radio, watch TV, or read newspapers and magazines
because of the content. Nearly everyone in the United States uses one
or more of these media every single day. While you are doing so, the
media blast ads at you. You can try to ignore those ads, but you cannot
be totally unaware of them. The Web, on the other hand, is not nearly
as essential to daily life, and probably never will be. There is content
there, but most of it duplicates content that can be found in other media.
Furthermore, it is easier to get the content from those other media. If
you turn on the TV or the radio, or pick up your newspaper from your
front door stoop in the morning, there is the content ready for you. You
can read or listen while you eat breakfast or while you are getting
dressed. You can listen in your car on your way to work. To access the
Web, you have to sit down at your computer. You cannot do this while
you are dressing, eating, or driving.
Once you get to work, however, the Web is available. Most employees
do not have TV or radio at work, but nearly every office worker today
has a PC, many of them with Web access. Office workers can, and do,
read and write email all day long. It has become essential to modern
office work. But email is not the same thing as the Internet. Employees
are supposed to be working while they are at work, not surfing the Net.
Some employees do listen to the radio while they work, but no one can
work while surfing the Net (unless they are in search of business information). So the Web is basically off limits for most people for most of
the working day. The same thing is true in the evening. You can sit down
at your computer in the evening, and millions do just that. But what most
people are doing is reading their email or playing computer games. Only
a few spend much time surfing commercial Web sites. It is an exhausting and boring experience after the first week or so.
So the Web cannot be an active medium, blasting ads at people. The
people are not watching. Web sites are passive; they are active only when
41
42
people decide to look at them for a particular reason, which does not
happen as often as people had at first assumed it would. What do people do on the Web? Some do personal (or professional) research. The
Web is the greatest research tool ever invented. You can find information on just about anything. Some visit chat rooms. Some get details on
news or events that go beyond what can be found in the mass media.
Some buy products that they cannot easily find in stores. Some listen to
or download music.
My vacuum cleaner story, I believe, is typical of consumer Web use.
I bought a Hoover with a microfiltration system that needed special bags.
When my bags ran out, I went to several supermarkets and department
stores in search of more microfiltration bags. None of them stocked the
bags. So in desperation I looked up Hoover on the Web. Sure enough,
the company had a Web site that offered the right bags. I purchased several. I am sure that Hoover does not make a profit by selling small parts
like this on the Web. What the Web site does for Hoover is make its
microfiltration vacuum cleaner a viable product. Without the Web, few
would ever be able to find the replacement bags.
The Web Is an Ordering Tool, Not a Selling Tool
In addition to vacuum cleaner bags, I have bought scores of books and
movies on the Web. I have bought vitamins and hard-to-get replacement
light bulbs and electronic parts. I would never buy clothing, an automobile, a TV, a house, a computer, or a microwave on the Web. Why not?
Because when I buy most things, I want to see and feel the product, or
talk to someone about it before I buy it.
Sears reports that many customers who are buying major appliances
have already done their research on the Web. When they come to the
store, they know what they want. But before they plunk down $1000,
they want to see the product and talk to a salesperson. I researched Dell,
Compaq, IBM, and Gateway on the Web before I bought the computer
I am using to write this book. I finally settled on Gateway. But when I
actually bought the computer, I called Gateway on the phone and talked
to a sales rep, who put the package together for me with an Iomega Zip
Selling on the Web
drive. This was a $1500 purchase. The sales rep talked me into a 1300megahertz model that was on sale, rather than the 1500-megahertz that
I had selected on the Web. I saved $300.
Web Advertising Was Highly Overrated
Since everyone assumed that consumers would flock to the Web and
spend hours every day looking at Web sites, it made sense to clutter their
computer screens with advertisements. A large percentage of the dot
com business plans included advertising as an important revenue source.
This was significant not only for the “portal” sites, such as Yahoo! and
America Online, but also for thousands of other sites, such as those of the
New York Times and the Washington Post and Women.com. Most of these
sites had no other source of revenue. At first, advertising agencies placed
substantial amounts of their clients’ ads on the Web, hoping for success.
When few people seemed to be responding to the ads, the idea was, “It
is early yet. As people get used to the Web, the sales will follow.” The
sales did follow in a few areas (books, travel, and financial services), but
even there, the profits were meager. The response rates to banner ads
dropped from 2 percent to tiny fractions of 1 percent.
After the crash, most major corporations drastically reduced their
advertising on the Web. The banners tended to be quick-decision,
low-budget operations. Gone were the shopping arcades and many
consumer product ads. Why? Because they didn’t work as well as mass
media ads.
Why Web Supermarkets Failed
Web supermarkets were featured in Don Peppers and Martha Rogers’
first book, The One to One Future. Peapod, a Web supermarket, was used
to explain how one-to-one marketing was the wave of the future. Unfortunately, Peapod is in trouble today, and most of its competitors have
failed. Why? The reason is not difficult to discover. Supermarkets
have become so competitive that their profit margin is only about 1
43
44
percent. That means that out of every dollar you spend there, the store
gets only a penny in profit.
A Web supermarket has to operate in that environment. It has to take
that penny and use it to pack customers’ orders in boxes (using either
low-cost workers or high-cost technology), put the boxes in refrigerated
trucks, and deliver them to houses scattered all over a metropolitan area.
No one can make any money doing that, and no one did. The Web supermarkets had to charge more than bricks-and-mortar supermarkets, but
even the extra charge did not pay for the service they were providing.
Furthermore, most Web supermarkets failed to achieve the profits
from the impulse buying that takes over when people visit a real supermarket. No one goes to a supermarket to buy a quart of milk and comes
home with just a quart of milk. But if you are ordering on the Web, you
haven’t made the investment to get into your car and drive to a supermarket, park, find a cart, wander down the aisles, and then wait in line at
the checkout counter. Everyone wants to make that investment worthwhile, so people pick up a few more items while they are there. A buyer
on the Web who needs only a quart of milk doesn’t see all the colorful
vegetables piled up at the entrance. She doesn’t see the ladies handing out
samples to eat. She doesn’t see the magazines at the checkout line. Her
only investment is the $10 delivery charge. To avoid that delivery charge,
she decides to pick up the milk at the 7-11 on her way home from work.
So, What Is Left? A Huge Research and
Transaction Tool
Consumers buy books, credit cards, travel, pornography, and financial
services on the Web. A much smaller number of people than had been
anticipated in the 1990s look at or respond to ads. What they often do is
research. Ordinary households use the Web as a combination of Yellow
Pages and Consumer Reports. They look up maps, cars, appliances, real
estate, and interest rates, and they use this information when they buy.
Because of the Web, consumers are far better informed today than they
were 5 years ago. It is not unusual to see a consumer going to a dealer carrying a PC printout that has all the background on what she plans to buy.
Selling on the Web
Dealers have to be more informed about their competition (and they find
it easier to be thus informed) because their customers are better informed
as a result of the Web.
My wife, Helena, and I were confronted with a problem. We had
bought a condo in Fort Lauderdale, but our two cars were in Arlington,
Virginia. We wanted to move one car to Florida and keep it there. I am
not up to driving to Florida, and neither is Helena. So what could we do?
We used www.Google.com and entered “Automobile Transportation.”
The first name on the list was www.intercitylines.com, which specializes
in moving cars in closed vans. The price was right, and we made a deal then
and there on the Web. How could we have found such a firm without
the Web? The company is in Massachusetts, and it does not advertise in
our Yellow Pages. This is what the Web is all about.
So what does this mean for those who want to use the Web to sell to
consumers today? Manufacturers and catalogers need to have a really
good site that has pictures and images of all their products and those
products’ features. They need to:
•
•
•
•
Make it easy for customers to look up their products.
Have a guide to local dealers by Zip code.
Use the Web to sell spare parts that dealers cannot or will not stock.
Make it easy to print on a PC what the consumer sees on the Web.
Premium Distribution
The Web has become a super method of distributing premiums or certificates. Here is how this works: Consumers arrive at commercial Web
sites to do their research. Many of them will print out a page containing
the information about your products. Since this is so, why not give them
a reason to purchase the product right away by giving them a certificate
(see Figure 3-1)? The certificate serves two purposes:
1. It drives traffic to the retailer.
2. The process of delivering it to the customer provides you with a
customer’s email name and address for further marketing efforts.
45
46
Here’s how to do this. When a consumer is looking at your product
on your Web site, there should always be a “Receive a valuable additional
premium” button on the site next to the item. When consumers click on
that button, they see a panel where they enter their name, address, and
email address. If they have already printed a certificate from your site
on this or a previous visit, the panel is automatically filled in, and all they
need to do is to click on the “Send Certificate” button. When they click
on this button, they will be sent an email containing the certificate as an
attachment, which they can print on their PC printer. Are you afraid
that some customers will abuse the certificates? No problem. You maintain a database, and if you allow only one certificate per product per
email address, your Web site can easily control that.
What the certificate contains are two bar codes. There is a bar code
for the product to be purchased and a bar code that has been created
for the customer. To use the certificate for a mail-in rebate, the customer takes it to a retailer, buys the product, and receives a purchase
receipt. He mails the certificate with the purchase receipt to the address
shown on the certificate. At the fulfillment center, the certificate is
scanned, capturing the product and customer data. There is no need for
Figure 3-1 Use of Web Certificates
Kodak Camera $345 for
$15 rebate Click Here
Consumer
Email certificate
Customer
& product
data sent
Buys
product
at store
Certificate mailed to
fulfillment;
rebate paid by check or
credit card
Selling on the Web
data entry at the fulfillment center, since your Web application automatically transmits all customer names to the center along with their bar
codes at the time they are sent the emailed certificate. The fulfillment
center simply checks the purchase receipt against the database created
by the Web site and clicks OK. If a rebate is involved, a check is issued
and mailed automatically.
A variation of this method, which can save the manufacturer money,
is to have the customer provide his credit card data, which can also be
converted to a bar code lookup table (not the actual credit card number) by the Web site software. At the fulfillment center, the scanning
process can credit the rebate to the customer’s credit card account, thus
saving the cost of mailing a check. The customer receives an email from
the fulfillment house simultaneously with the mailing of a check or the
crediting of the credit card.
There are several important elements in this process:
• The customer cannot provide a fake email address, since a valid email
address is required in order to get the certificate. She can’t provide a
fake name or address, or she will not get a cashable check. She can’t
provide a fake credit card number, or she won’t get a refund.
• In applying for the rebate, the customer gives permission for the use
of her email address for future promotions. If she does not want to
receive information, the rebate could be smaller or not available.
• The manufacturer gains a database of customers and email
addresses, which can be used for promotion.
• If the customer does buy the product, that fact is stored in the
database and makes the database more valuable.
• The fulfillment house process is fully automated and therefore is
cheaper and faster.
• The manufacturer can get daily reports on the success of the
rebate program by product, state, region, and type of consumer.
• When registering for the certificate, the consumer can be asked
for more information, such as income, age, and so on.
The certificates should not offer discounts. Discounts erode value. A
better offer would be a premium, such as training or software that
would go with the camera, or a rebate.
47
48
Pure-Play Selling on the Web
Less than 5 percent of the people who visit an online retail site actually
buy something. Compare this with the performance of someone who
drives up and parks at Wal-Mart or a supermarket. No one walks away
from this trip empty-handed. Why don’t people buy on the Web? Many
sites are still too difficult to navigate. Flashy graphics that take minutes
to download turn people off. The rigidity of shopping online is frustrating. My experience is typical.
When Orbitz.com opened its travel site, I was one of the first to use
it. I ordered a cheap Thanksgiving flight to Manchester, New Hampshire, for Helena and me. Then I called my son David, who said that
the timing was wrong. Orbitz.com would not let me change my flight
times, despite several emails. I ended up wasting $400. My travel agent
has always been able to make changes in nonrefundable flights. The
irony of the situation is that when Thanksgiving rolled around, I
received a maddening email from Orbitz.com telling me to get ready for
my wonderful trip to Manchester. I will never again use Orbitz.com
for anything. Its inflexible method of operation turned me off permanently. Its customer service was totally unresponsive. This is not what
the Web should be.
Brooks: Clever Efforts to Promote Web Sales
To keep more customers, Web retailers have had to change their ways. To
attract and hold customers, Brooks used a service called the Angara
Reporter that collected demographic information, such as age, gender, and
location, from At Home Corp.’s MatchLogic.com and Engage Inc.
visitor profiles to create a rough idea of the kinds of customers who visit
particular types of sites. Brooks used this information to personalize the
content that visitors saw and to tailor offers on the basis of the profiles.
Angara’s profile was based on a database of 100 million Web visitors.
Visitors to Brooks’s site were sorted by the site’s best guess as to
gender, marital status, and geography. Different visitors saw different
home pages. A woman visitor would see a different page from that
Selling on the Web
shown to a man. An unmarried younger man might see a different page
from that shown to an older male. Brooks expected that this feature
would offer a level of personalization that its regular stores could not
match. Yet, as of this writing, Brooks has yet to make a profit. What
can we learn from this?
• There are many very sophisticated and clever things that you can
do using a database to create sales to consumers on the Web that
you could not possibly do in regular merchandising.
• These things all cost money.
• Total Web sales are very tiny in comparison to retail sales.
• These clever things have yet to produce a lasting profit for anyone.
• Anyone trying to sell consumer products on the Web (with a few
exceptions, such as Amazon and eBay) is going to lose money for
the foreseeable future.
Dealer Locator
Commercial Web sites should make it very easy for customers to find
dealers. This sounds like an obvious suggestion, but in fact it is often
very difficult for manufacturers to provide a dealer locator. Most goods
are sold through wholesalers or distributors. These folks keep the dealers’
names to themselves. Yet providing information to the ultimate consumer
is a very efficient way of delivering customers to dealers and building
total market share.
How can you find dealers? Follow the money trail. Make it easy and
profitable for dealers to be identified. To do this, have a Web microsite
where dealers can provide their location, phone number, directions, store
hours, and other such information (Figure 3-2). When dealers come to
your microsite to register, have them provide the name of the wholesaler
or distributor that supplies your products to them. Once your Web site
has captured the dealers’ name, the Web site can automatically send the
information to the appropriate wholesaler or distributor for verification
before it becomes a part of your consumer Web site dealer locator. Some
wholesalers or distributors will help you in this process; others will not.
49
50
Figure 3-2 Finding a Dealer
Dealer enters
info on your
microsite
Microsite for
dealer
registration
Your
consumer
Web site
Your database
Wholesaler
OKs dealer
entry
The dealers provide you with their information plus their email
address. Then, when customers come to your Web site and ask for a
local dealer, if the customers give you permission, you can send their
names to the dealer for follow-up.
Citigroup Strikes Out
In 1997, Citigroup, the parent company of Citibank, launched e-Citi,
a Web bank that was designed to compete with Citibank itself and other
parts of the $230 billion company. Citigroup put a lot of resources into
the venture. Within 3 years the Web bank had 1600 employees and 100
different Web sites. Between 1998 and 2000, Citigroup spent $1.1 billion
on the effort. It was a gigantic failure. By March of 2000, when Citigroup
called it to a halt, the venture had only 30,000 accounts, compared to
the 146 million accounts in Citibank.
What went wrong? For one thing, e-Citi was set up as an independent
venture. Customers could not use Citibank branches. This was a fundamental mistake. People using a Web bank have a large number of
questions to start with:
• How do I deposit money?
• What ATMs can I use?
Selling on the Web
• Whom do I talk to when I have a problem or need a loan?
• Whom can I relate to?
Since, on top of all that, there was no institution to identify with, it
is amazing that Citigroup was able to find 30,000 people who were willing to open accounts.
No Web bank has ever made a profit, and none ever will. What can
be profitable is for an existing bank with lots of branches to persuade
some of its customers to use the Web for some of their transactions.
Web transactions cost about one penny each. Branch visits cost at least $2.
The cost savings to the bank can be very substantial. But customers will
use the Web only if they have confidence in the institution and if there
is someone whom they can go to see when they need help.
After Citibank folded the e-Citi operation, the company formed an
Internet Operating Group to see which Citibank services could be profitably shifted to the Web. The company created an online-payments
business called C2It that let customers email money to one another for a
1 percent fee. Individual Citibank units were encouraged to shift functions
to the Web. Eighteen months later, Citigroup was serving 10 million customers online with a variety of services. Looking forward, Citigroup
expects to cut $1 billion off its annual costs through using the Web.
eToys Strikes Out
When eToys went public in 1999, its market value reached $7.7 billion.
The company was backed by Intel, Sequoia Capital, and Idealab. What
a great beginning! Its Web site included a sophisticated search engine
that let parents look for toys appropriate for a child of a particular age,
or for toys of a particular brand or theme. It created “wish lists” that let
children choose gifts and email their lists to family members. Once the
firm’s Web site was out there for everyone to see, however, competitors
like Amazon and Wal-Mart were in a position to copy the firm’s ideas
and beat its prices, and they did. Soon competitors began to sprout up
everywhere: KBKids, Smarterkids, Toysmart, ZanyBrainy, NuttyPutty,
JC Penney, Target, and FAO Schwarz joined Amazon and Wal-Mart in
selling toys on the Web. Many of them spent lavishly on advertising.
51
52
eToys assumed that loyal customers would come back once they had
been acquired. On the Web, customers soon learn that it is very easy to
shop around—and they do. Proximity works with bricks-and-mortar
stores. You can drive to them, park in their parking lots, and go inside.
The Web is very different. It is very hard to create and maintain customer loyalty on the Web. eToys could not count on its customers
coming back, and they didn’t.
Even given the competition, eToys assumed that the growth of the
Web would provide enough buyers for everyone. The company planned
to double its sales every year. That was achievable only when sales were
very small. In fact, competitors multiplied and national sales grew only
modestly, and so eToys’ forecasts were worthless.
Perhaps eToys’ biggest mistake was spending too much to create a
distribution network. The company’s investment in property and equipment amounted to more than $120 million. This meant that property
investment equaled about 95 percent of the year’s revenue. A similar
figure for retailers was 20 percent, and one for catalogers was 13 percent.
Faced with these investments, eToys could not make a profit without a
massive sales increase. That increase did not occur.
Business-to-Business Web Transactions
The Web has proved to be an excellent transaction medium for business-to-business dealings. Some companies selling memory chips for
computers, for example, have registered sales of several hundred million dollars per year. For a few, these transactions represent more than
85 percent of their total revenue. Their sales are to companies that buy
computer components to install in units for sale or to upgrade their own
equipment. This type of Web transaction will continue to grow in the
years ahead.
How do customers find out about these transaction sites? About a third
of them comes from advertisements placed in print magazines, a third
comes from Web ads on technical sites, and the remainder comes from
search engines. Many have established premier pages for their best customers. On these pages, their best customers have established accounts,
Selling on the Web
so that their employees can order the units charged directly to their company accounts.
Business-to-business Web transactions continue to register success in
a wide variety of fields. Why are they more successful than consumer
Web sales? There are several reasons:
• The average order size of business transactions is much greater
than the average consumer order size.
• The universe of business customers is much smaller. It is less
expensive to reach them through direct mail or print or Web
advertising than to reach average consumers.
• Premier pages work well here. They don’t work at all with
consumers.
• Business customers have a continuing and predictable need for
most of the business products sold through the Web. Consumer
purchases cannot be predicted as easily.
• Business-to-business transactions on the Web are conducted
during the day, when buyers and sellers are in their offices
working at their company PCs. Consumers, on the other hand, if
they are employed, have to do their Web shopping after they get
home at night. There are competing demands on their time. It is
more fun to go to the mall with their families or to watch
television. That is why about 80 percent of all sales on the Web
are business to business.
Selling on the Web: What We Have Learned
• Business-to-business e-commerce transactions can be highly successful.
• Pure-play Web e-commerce seldom works. Success comes to established
businesses that add a Web presence. To make the Web work, you
have to build a very expensive distribution system. Once you have
built such a system, you are not a pure-play operation anymore.
• Barriers to entry on the Web are very low. It is very easy and
inexpensive to start a new Web site in any category. It is very hard
and expensive to build a chain of stores and warehouses. That
53
54
means that if you are successful on the Web, you will very soon
have many Web competitors.
• People like to hold something in their hands before they buy it. What
about catalogs? A catalog is something you can hold in your hands,
pass around, and scribble on; you can tear out pages and stick them
on your refrigerator. You can’t do any of this easily on the Web.
• There is precious little loyalty on the Web. To keep people coming
back, you have to personalize your Web site and communicate
often. Mass advertising is too expensive for most Web sites.
Loyalty is very difficult to generate and maintain on the Web.
So if you have an established land-based business and want to benefit from the Web, what should you do?
• Look on the Web as more of an ordering medium than a selling medium.
Put your Web URL on all your catalogs, brochures, and direct mail
pieces. Provide one-click ordering to make it super fast.
• Provide a dealer locator.
• Create a profile of every customer. Give all your customers, whether
business-to-business or business-to-consumer, a personal PIN, and
invite them to come to your Web site to update their profile. Give
them a gift for doing so. What is a profile? It contains not just the
customer’s name and address, but the customer’s email address (with
permission to use it), demographics, family composition, preferences,
and purchasing history. Get the customer’s permission for you to
remember his or her ID and password—that is, permission to put a
cookie on his or her PC. If the customer is a business, use the profile
to find out what it does, how it uses your products, and how often it
buys. Use the profile to populate your database that is used for
regular offline sales.
• Use the profile to personalize your site. As Amazon or Staples does,
say, “Welcome back, Susan” (using a cookie). Then vary the site
content based on the profile. If Susan has children under the age
of 4, your site will look different from the way it will look if she
has children in college. This is not expensive to do. Most sites that
have customer profiles developed can become fully personalized
for a hundred thousand dollars—not millions.
Selling on the Web
• Communicate often. When customers hit the one-click order
button, thank them then and there for the order and reiterate what
it was for. While they are reading this, send them an email
repeating your thanks. Then send an email saying, “Your order
will be shipped on . . .” Then another email: “Your order was
shipped by UPS with this tracking number . . .” And another
email a week later: “Did your order arrive on time? Were you
satisfied with it? Click Here to let us know your reaction.”
Success on the Web
If you can’t make money selling on the Web, what should you do? Measure your success in other ways. This should not be total sales. Total
Web sales are still a small fraction of all sales. For most catalogers, Web
sales seldom exceed 15 percent of total sales. What the Web does is
permit you to save some money, if you do it right. You have to have a
Web site today. It is how companies are judged by customers, suppliers,
and investors. You can’t hide your head in the sand just because you cannot make money from your site. So build it, and keep the costs down.
Then do the following things:
• Put lots of free information on the Web. Every company has information
lying around that could be made available to customers. Magazines
and newspapers have archives. Technical companies have lots of
technical reference data. Travel services (hotels, resorts, airlines,
travel agencies, cruise lines) have massive amounts of information
about destinations: maps, weather, activities, amenities, local lore and
history, tours, and so on. Put all that on the Web for free. Why?
Because it will save you lots of money. Encourage travelers to find
out what it is like to go to Italy by reading your Web site and looking
at pictures rather than by talking to your customer service reps all
day. Once they have made up their minds, they can then call a CSR
to book their trip.
• Measure your success by reduction in calls to CSRs. Begin the year by
estimating the number of incoming and outgoing calls that you
expect to field during the year. Analyze those calls. How much
will they cost? Why are they made? Which types of call could be
55
56
•
•
•
•
handled on the Web? Then set up your Web site to answer the
questions most likely to be asked of a CSR. “Where is my order?”
“How do I get replacement parts?” “What is the weather going to
be like in Prague?” If you do this right, you will be able to save
millions of dollars by shifting these calls to the Web. I worked with
a travel-related firm that received more than 7 million telephone
calls a year, costing it more than $48 million. This company
measured its Web success by the number of these inquiries that it
could get its customers to make on the Web. Each CSR call cost
the company $6.50. Each Web visit cost less than $0.10. If even 10
percent of these calls could be shifted to the Web, the company
could easily save millions of dollars. Some companies that have
done this find that CSR calls do not decrease, but that customers
use the Web to get the information and call the CSR to close the
deal. This is wonderful!
Sell parts and replacements. Any established business can profitably
provide replacement parts through the Web. Look at Hoover.com,
where you can find replacement parts that are not carried in any
retail store. Go to GEAppliances.com, where you can find 50,000
parts that you can buy on the Web.
Travel sites can help your travel business. Travel-related businesses can
profit from being on the Web even if they don’t sell anything
directly. Hotels, car rental agencies, airlines, and amusement and
theme parks can benefit. I am writing this from Fort Lauderdale,
Florida, where I am spending some time with my children and
grandchildren. We needed a six-passenger minivan, but Hertz, Avis,
Budget, National, etc. were sold out. I found a brand new minivan
from Payless Auto Rental (who ever heard of that firm?) on the
Web. How else would I have been able to locate such a company?
Publicity. Every rock star today has a personal Web site. Why?
Vanity? Not at all. If you register at a Web site for Sheryl Crow
or Madonna, you will be added to a list. When the singer’s next
album is released, you will probably get an email. Experts in the
industry can demonstrate a relationship between emails to fans
and retail store sales. Publicity on the Web can be made to pay.
Sell to business customers. This is a major growth area.
Selling on the Web
Why Selling on the Web Cannot Be Easy
Using the Web to compete with established businesses was a pipe
dream. It could never work. Why not?
• Established businesses spent millions to get where they are, and
they will not fall down and play dead just because some Web site
comes along.
• If it were easy, you would have lots of Web competitors, and you
would soon find it difficult.
• Furthermore, if it were easy, the established businesses could always
beat you by doing exactly what you are trying to do, because they
know a lot more about it than you do.
Measuring the Value of a Site
From the previous discussion, it is obvious that for most companies,
Web sites cannot be successful sales channels to consumers for most of
their products. They are primarily ordering vehicles and information
providers. They may be excellent sales channels for spare parts and
supplies and for obscure and slow-moving items that dealers will not
stock. To think properly about the role of the Web site, we must begin
to think of it as a cross between mass media ads, spare parts distributors,
and customer service. The value of a manufacturer’s consumer Web site
today is estimated in Table 3-1.
Table 3-1 Monthly Value of a Commercial Consumer
Web Site
Visits per month
Customers registered
CSR time saved
Data printed
Rebate certificates printed
Certificates redeemed
Parts sold
Total value
Customers
Rate
Value
500,000
20,000
50,000
40,000
25,000
15,000
8,000
$ 0.02
$ 1.00
$ 4.00
$ 2.00
$ 2.00
$64.00
$12.00
$ 10,000
$ 20,000
$ 200,000
$ 80,000
$ 50,000
$ 960,000
$ 96,000
$1,416,000
57
58
Let’s start one step up from the bottom:
• Parts sold. Every Web site should sell spare parts and replacement
items that dealers will not or cannot stock. These parts are not a
profit center. They are a reason for consumers to visit the site.
They can be the most important draw on any site and can turn
existing customers into advocates. To determine the value of parts
sold, we have to consider three factors:
• The profit made on the parts themselves, which may be
quite low, depending on how you want to price them
• The acquisition of a customer name and email address,
which can be valuable for marketing
• The goodwill and brand name publicity value of the site
• Certificates redeemed. To determine the value of these certificates,
we need to consider the profit from the sale of the item less the
rebate and fulfillment costs. If the profit, for example, is $80, we
must subtract $15 for the certificate and $2 for the fulfillment
cost, giving us $63 net profit. To be totally accurate, you can make
the assumption that some of your products might have been sold
to these same people without the certificate, so that the profit
from the sale that is attributable to the Web would be less than the
full profit. You can experiment and test to get the right answer.
• Certificates printed (whether redeemed or not). There are two benefits
of having a certificate printed. First, of course, is the fact that
some of the certificates will be redeemed, which has already been
discussed. But whether a Web certificate is redeemed or not, it has
value to the issuer, because before the certificate is printed, we
require a valid name, address, and email name. To determine the
value of a name, we can use the method described in Chapter 5,
“The Value of a Name.” Here we are assuming that it is $2.
• Data printed. When people go to your Web site, you make it easy
for them to find information and print it out on their PC printers.
Make sure your name and identification is on everything that they
print. People who are looking for a refrigerator may print out
information from your site and take it with them when they visit a
store. They will print out the specifications of your automobiles,
Selling on the Web
•
•
•
•
your swimming pools, and your condominiums. This is wonderful!
It’s great advertising to have detailed information on your products
in the hands of people who are about to make a purchase. This is
worth a lot of money to you, and it is much cheaper than mass
marketing, where you pay for ads that millions will not read or
listen to.
CSR time saved. One important function of your Web site must be
to reduce the time that your customer service reps spend explaining
things that are more easily read and printed by customers. Put
thousands of pages of information on your site free. Provide a
super search engine so that anyone can find anything in three
clicks. Measure your success by hours of time saved by customer
service reps at $4 per telephone call. Encourage your CSRs to get
customers to do the research themselves.
Customers registered. Of course you want customers to register if
they come to the site. Most of them won’t, of course, but many
will. You can follow up with an email thanking them for the site
visit and answering any questions that they posed when they
registered. Why is registration worth money? Because you will
use the email names that they give you to send marketing
messages to in the future.
Web visits. There is value to a Web visit, whether or not the
consumers who visit print data or certificates. Your Web page is,
after all, an advertisement. We can assign a value to this, just as
we do to TV, radio, or print ads. We are assuming that a Web
visit is worth $0.02.
Total value. Assigning a total value to a Web site is important. It
permits you to decide how much to spend on the Web site, and
it lets you budget for the staff that is maintaining it.
If you do a good job of designing your Web site, it will capture many
customer names every month. These names may be used for marketing
programs. In this example, the total value is $1.4 million per month.
Your site may not be as valuable as this, but you can use this table to
estimate the benefit of your site to your company. It is probably worth
more to you than you thought.
59
60
What Works
• Business-to-business Web transactions.
• Using the Web as an ordering site to support paper catalogs.
• Using the Web to sell spare parts and replacements for products
sold in retail stores.
• Using the Web as a dealer locator.
• Using the Web to capture information about customers that can
be used to create emails that will drive those customers to your
site to make a purchase.
• Providing free information and pictures on your Web site to save
the time of customer service reps.
• Using the Web to replace customer service time.
• Distributing certificates through the Web.
• Registering customers on the Web.
What Doesn’t Work
• Thinking of the Web as a sales medium. It isn’t. It is an ordering
medium.
• A pure-play, stand-alone, consumer selling Web site (with limited
exceptions for travel and pornography) without a catalog or retail
stores.
• Assuming that advertising on the Web will be the main source of
revenue.
• Catalogs on the Web without paper catalogs to drive people to
the site.
• Web supermarkets.
Quiz
1. Why did the judge’s decision in the Microsoft case lead to the
dot com crash?
a. Microsoft was the heart of the Web.
Selling on the Web
2.
3.
4.
5.
6.
b. The public feared that the government was going to go
after the Web next.
c. Profits on the Web were very high.
d. The Web was a major sales medium.
Why did Web advertising fail?
a. Ad prices were too high.
b. There were too many ads on the Web.
c. The Web is a passive medium.
d. Competing media combined to kill it.
e. Few Web sites depended on ads for their revenue.
Why did all Web supermarkets fail?
a. Because no one wanted to have groceries delivered
b. Because of the economics of the supermarket industry
c. Because major chains fought them
d. Because investors could not be found to start such
enterprises
e. Because selling groceries on the Web was a crazy idea
What are the flaws in the Web certificate idea?
a. There are none. It is a wonderful idea.
b. Customers will give fake email names.
c. Customers will give fake addresses.
d. Customers will abuse the system by sending in for too
many refund checks.
e. Customers hate certificates and will not use them.
Why did e-Citi fail?
a. Citibank did not give it adequate funding or time.
b. The e-Citi customers could not use Citibank branches.
c. Citibank was not very well known.
d. Citibank did not devote enough employees to the project.
e. The reason is unknown, since most Web banks are highly
successful.
Why did eToys fail?
a. It had inadequate market capitalization.
b. Its site was not very sophisticated.
c. Many competitors copied its ideas and stole its customers.
61
62
7.
8.
9.
10.
d. It did not spend enough on its distribution network.
e. The answer is unknown, since most Web-only merchants
have been successful.
Which was not suggested as a way for land-based businesses to
be successful on the Web?
a. Provide a dealer locator.
b. Get customers to provide profiles.
c. Keep customer communications to a minimum.
d. Provide free information on your site.
e. Personalize your site using cookies and customer profiles.
Which was not suggested as a way to succeed on the Web?
a. Provide free information.
b. Eliminate telephone customer service.
c. Sell replacements and parts.
d. Have a Web site.
e. Encourage publicity.
Which does the chapter suggest is the most profitable single
activity on the Web for a typical company?
a. Selling replacement parts
b. Advertising on the site
c. Customer visits
d. Certificates
e. Customer service
Which is the one reason not given for significant success from a
Web site?
a. Customer visits
b. Advertising
c. Publicity
d. Customer service
e. Consumer product sales
C
H A P T E R
4
Computing Lifetime
Value
I think that without numbers there is no future. You cannot measure
your progress unless you’ve got benchmarks against which to evaluate
costs and profits. So ROI, RFM, and lifetime value are the minimal
critical benchmarks you must now establish if you want to measure the
profitability of your business today. Those same benchmarks then
become the standards by which you measure your business’s future
growth and the justification for the database tools you must have to get
that growth. Without the numbers, you don’t have a future.
—John Travis, Manager, Business Development, Hudson’s Bay Company
T
here are really only three marketing functions:
1. Acquiring customers
2. Retaining or reactivating them
3. Selling them more products
Everything else is a modification of one of these three functions.
Measuring customer lifetime value is central to all three functions, as
we will see in this chapter. Lifetime value tells you how much you
should be spending on each function, and it measures whether your
marketing efforts are doing any good.
63
64
Lifetime value is a measure of the net profit that you will receive from a
given customer during his or her future lifetime as your customer.
I have computed the lifetime value of customers for a great many
different companies, including banks, credit card issuers, insurance companies, packaged goods companies, software companies, hardware
companies, and automobile companies. The case study given in this
chapter deals with a fictitious automobile company. It is interesting
because it illustrates how to compute the automobile repurchase cycle
over a 9-year period, which is what automobile companies have to deal
with. The case illustrates the methods used. The numbers have been
modified so that they are not derived from any particular automobile
company—the case does not disclose any proprietary information, but
rather is a summary of my experience with three very different automobile companies over several years.
Asian Automobiles
In this chapter, we will take you through the computation of lifetime
value for an automobile company, which we call Asian. Then we will
show you the strategies that arise as a result.
Asian sells automobiles and arranges car leases and loans. The company had 2 million customers on its database. These customers included
current owners of new Asian cars, owners of used Asian cars, and Asian
leaseholders. The first and most important part of the lifetime value
project was determining the repurchase rate of Asian’s customers.
The repurchase rate for automobiles is very complicated. Some
people buy a new car every year. Others buy them only every 2, 3, 4,
5, 6, 7, or even more years. Some buy two or more cars. Others buy
used cars. Still others buy an Asian, drive it for a while, then sell it and
buy some other brand. Then they come back to Asian after a few years.
In all of this buying and selling, what is the repurchase rate?
The repurchase rate for automobiles is the percentage of the average group
of customers who will buy a new car from the same manufacturer when they
turn in their existing vehicle.
C o m p u t i n g L i f e t i m e Va l u e
The repurchase rate is further complicated by the fact that Asian has
many different models of cars, and each model has its own unique
repurchase rate.
The goal for any automobile company is to have a repurchase rate of
more than 100 percent. How would that be possible? It would occur if
customers never died, if they always bought the same brand when they
turned in their existing vehicle, and if many of them bought two or
more cars. For the automobile industry as a whole, the repurchase rate
is about 35 percent. The goal is to determine the repurchase rate for
every Asian model and for owners of used Asian cars as well.
To accomplish this, it is necessary to develop some tables from the
database that illustrate the situation. Such tables should begin with a
particular year, such as 1997. You can then determine what Asian cars
the owners of 1997 models owned in the summer of 2002. Next, you
create the same tables for the owners of 1998, 1999, 2000, and 2001
models. Finally, you have to create similar tables for Certified Previously
Owned (CPO) Asian cars—Asian cars that were bought used. These
data are usually available because manufacturers require their dealers to
report the owners of each new and each used Asian. This information is
stored in Asian’s database. Table 4-1 shows what the 1997 Asian owners
owned in the year 2002.
From these data, one would assume that the repurchase rate varied
from 22 to 38 percent, but that is not the case. To get the real repurchase rates, you have to figure in the Asian owners who bought cars
every year, every 2 years, and so on. The numbers shown in Table 4-1
mask the fact that in the 5 years between 1997 and 2002, a significant
Table 4-1 Changes in Ownership over the Years
Chart A
Totals
Owned 1997
120,409
Owned 2002
91,875
Bought new
30,295
A Model
9,491
B Model
3,735
C Model
8,491
D Model
6,684
E Model
1,893
Repurchase rate
25%
A Model
B Model
C Model
D Model
E Model
17,472
14,133
4,003
980
1,729
651
396
247
23%
53,259
38,950
11,666
6,221
1,040
3,047
838
521
22%
26,784
19,585
7,861
1,464
565
3,924
1,267
641
29%
17,646
14,111
6,619
802
382
844
4,126
466
38%
452
339
146
24
20
26
57
19
32%
65
66
number of people bought two, three, or more new cars from the company and resold them, and many purchased multiple vehicles and kept
them. Dozens of charts are needed to measure and account for the varying behavior of customers. Overall, the picture that emerged was of the
average Asian customer buying a new car every 3 years because of the
expiration of the lease and loan arrangements. The raw numbers have
to be converted into adjusted numbers as shown in Table 4-2.
Putting the numbers from all years together gives us the results
shown in Table 4-3. From Table 4-3, we are able to go on to compute
the automobile owners’ lifetime value. The lifetime value chart for the
B Model is shown in Table 4-4.
Table 4-4 needs some explanation. In the first place, the data showed
that the repurchase rate tended to go up for long-term Asian car owners.
Long-term owners tend to be more loyal than newer owners. The
$4000 profit per car is the profit to Asian, not the profit to the Asian
dealer. Asian is actually selling cars to dealers, not to the general public.
The acquisition cost per car was the amount per car that Asian spent on
co-op ads and other promotions. Each dealer, of course, had its own
Table 4-2 Adjusted Repurchase Rates
Totals
A Model
25.16%
24.00%
16.00%
7.00%
4.00%
22.91%
16.00%
11.00%
5.00%
4.00%
7.90%
11.50%
15.80%
7.00%
7.90%
0.10%
7.40%
8.20%
10.90%
5.20%
7.20%
39.00%
B Model
C Model
D Model
E Model
29.35%
26.00%
14.00%
6.00%
4.00%
37.51%
35.00%
23.00%
8.00%
5.00%
32.28%
17.00%
20.00%
20.00%
20.00%
6.00%
11.30%
16.10%
5.60%
6.10%
45.20%
11.00%
13.10%
14.10%
6.40%
7.10%
51.60%
12.70%
17.40%
22.60%
8.00%
9.90%
70.60%
10.90%
8.50%
20.00%
20.00%
40.00%
71.00%
Raw Totals
1997
1998
1999
2000
2001
21.90%
23.00%
16.00%
6.00%
3.00%
Adjusted Totals
1997
1998
1999
2000
2001
Total
0.33
0.5
1
1
2
5
Table 4-3 Final Repurchase Rates
Repurchase rates
Totals
A Model
B Model
C Model
D Model
E Model
50.10%
39.00%
45.20%
51.60%
70.60%
71.00%
Table 4-4 Lifetime Value
B Model cars
Revenue
Repurchase rate
Customers
Net profit per car
Net revenue
Years 1–3
45.20%
264,124
$4,000
$1,056,496,000
Years 4–6
Years 7–9
47.40%
119,384
$4,000
$477,536,192
51.20%
56,588
$4,000
$226,352,155
Costs
Acquisition cost per car
Acquisition cost
Marketing costs ($20/yr)
Total marketing and sales
$400
$105,649,600
$ 5,282,480
$110,932,080
$
$
$
Profit
Discount rate
NPV of profit
Cumulative NPV of profit
Lifetime value
$945,563,920
1.11
$851,859,387
$851,859,387
$3,225.23
$ 439,333,297
1.31
$ 335,368,929
$1,187,228,316
$44,94.97
$300
35,815,214
2,387,681
38,202,895
$
$
$
$275
15,561,711
1,131,761
16,693,471
$ 209,658,684
1.54
$ 136,142,002
$1,323,370,318
$50,10.41
advertising and promotion costs, which are not reflected in these numbers. The marketing costs ($20 per customer per year) are the amounts
that Asian spent annually on the database and on communications with
customers. This is a high number. Many car companies spend only a
fraction of this amount on marketing to existing customers.
The profit is equal to the revenue less the costs. This profit has to be
discounted to reflect the net present value of the funds involved.
The net present value is the current (today’s) value of money that you are
going to receive in the future, as $1000 to be received a year from now is
not worth as much as $1000 received today. To compute the value of
$1000 to be received a year from now, you have to discount it by an
amount that reflects the cost of waiting. Waiting has a cost, because if
you had the $1000 now, you could invest it and receive a return. By
waiting, you forgo that return. We discount all future profits in a lifetime value chart to produce net present value. The numbers behind
Table 4-4 are shown in Table 4-5.
The discount rate reflects the fact that for Asian, automobile sales
are a business-to-business transaction between Asian and its dealers. See
Chapter 2 for the formula for the discount rates.
Finally, of course, the lifetime value of a customer is derived by
dividing the cumulative net present value number in each of the three
67
68
Table 4-5 Discount Rate
Conversions
Factor
Rate
Time delay 1–3 years
Risk factor
Cost of capital
2.0
5.0
8.0
1.20
4.6%
time periods by the original 264,124 customers. What we are computing is the lifetime value of these 264,124 customers. Table 4-4 tells
us that the value of a newly acquired customer to Asian is $5010 in
the ninth year.
What You Can Do with the Numbers
Knowing what the lifetime value of a customer will be 9 years from now
is not very valuable unless you can use that number to drive marketing
strategy in some way that will increase company profits. That is exactly
what we are going to do. Let’s look again at the repurchase rate by
model (see Table 4-3).
The average Asian repurchase rate is about 50 percent, but the repurchase rate for B Model cars is only 45.2 percent. The B Model is Asian’s
most popular car. Suppose the repurchase rate for this car could be
raised so that it was equal to the average repurchase rate for all Asian
cars, namely 50 percent. This is certainly an achievable goal. If it were
to be reached, what would it do for Asian’s profits? We can calculate that
by making two assumptions:
1. We can come up with strategies that raise the repurchase rate of
this model of car.
2. The strategies will succeed in bringing the repurchase rate up to
50 percent, or about even with the repurchase rate for the average
Asian car.
What might these strategies be? There are many things that Asian
might offer its customers:
• A welcome kit inviting new owners to come to a Web site to
update their profile and to provide an email address for future
correspondence
• Regular communications about the car, including service
requirements
• A car birthday card
• An Asian magazine for new-car owners only
• Invitations to events sponsored by Asian, including skiing and
yachting excursions, golf tournaments, and fashion shows
• Special preview information for owners on new Asian models and
features
• A customer service call twice a year:
“Welcome, Arthur. I am Jennifer, your personal Asian customer
service representative. I am here to make your B Model ownership
a wonderful experience. Here is my personal phone number. Call
me whenever you need assistance of any kind. You can reach me at
any time. I am here to help.”
What could Jennifer do?
• She could provide the owner with the distance and
directions to the nearest gasoline station, hotel, restaurant,
or hospital.
• She could call the police, when needed, or provide
emergency road service.
• She could help to locate cars that have been stolen or that
have been lost in a vast airport parking lot.
• She could locate the nearest Asian dealer and schedule service.
• She could communicate with the owner by email, sending
birthday cards and service reminders.
In other words, it is possible to make owning an Asian B Model so
attractive that enough owners will want the same model next time to
make the repurchase rate climb from 45 percent to 50 percent. Let’s
say that the cost of these additional services would be $30 per car per
year. Putting that increase into the calculations, Table 4-6 shows what
would happen to B Model lifetime value. Comparing before and after,
we have increased Asian’s profits by $56 million over 9 years, as shown
in Table 4-7.
69
70
Table 4-6 Lifetime Value after Retention Programs
New LTV
Revenue
Repurchase rate
Customers
Net profit per car
Net revenue
Years 1–3
Years 4–6
Years 7–9
50.00%
264,124
$4,000
$1,056,496,000
53.00%
132,062
$4,000
$528,248,000
56.00%
69,993
$4,000
$279,971,440
Costs
Acquisition cost
Total marketing & sales
$400
$105,649,600
$ 13,206,200
$118,855,800
$300
$39,618,600
$6,603,100
$46,221,700
$275
$19,248,037
$3,499,643
$22,747,680
Profit
Discount rate
NPV of profit
Lifetime value
$937,640,200
1.11
$844,720,901
$844,720,901
$3,198.20
$482,026,300
1.31
$367,959,008
$1,212,679,909
$4,591.33
$257,223,761
1.54
$167,028,416
$1,379,708,324
$5,223.71
Years 1–3
Years 4–6
Years 7–9
$3,198.20
$3,225.23
($27.03)
($7,138,486)
$4,591.33
$4,494.97
$96.36
$25,451,592
$5,223.71
$5,010.41
$213.30
$56,338,006
Table 4-7 LTV Gain
New lifetime value
Previous lifetime value
Difference
Times 264,124 customers
Let’s see just what this $56 million profit means. If you look at years
1 through 3, you will see a loss of $7 million. The loss comes about, of
course, because we are spending $30 more per customer per year to
build and maintain a profitable relationship with Asian’s customers.
These improved relationships will not show up until the 3 years are over
and Asian’s customers make a decision on their next new car. Without
a long-range plan, few car companies would be willing to stand for a $7
million loss in the first 3 years. Numbers like this are the reason why
database marketing is such a hard sell for all but the most established
and forward-thinking companies. But the $56 million profit is real. It
includes all costs, and it absorbs the $7 million loss in the first 3 years.
As we shall see later, this $7 million loss is not real. It is offset by gains
from buyers who get a new car every 1 or 2 years.
Buying a Car Every Year
Relationship building does not need to stop with moving the B Model
repurchase rate up to 50 percent. In fact, there is much more that you
could do with the lifetime value analysis in this situation.
The tables given so far have assumed that the average Asian customer
buys or leases a new car every 3 years—and, in fact, that is the case.
What it ignores, however, is the fact that there are a substantial number
of Asian customers who buy new cars every year or every 2 years. What
will be the impact of the retention-building programs on these people?
Detailed studies of the Asian database showed the following:
• Six percent of Asian B Model owners bought a new car every year.
• Six percent of Asian B Model owners bought a new car every 2 years.
• The overall increase in sales due to the new program will affect not
just the third-year totals, but the interim-year totals as well, because
your communication strategies will affect these 1- and 2-year buyers
and get them to improve their repurchase rates.
There is an extra 1.2 percent increase in sales in addition to the overall 5 percent increase (from 45 to 50 percent), for a total increase of 6.2
percent. Plugging that 6.2 percent increase into the tables raises the $56
million to $73 million. But that’s not all.
Automobile Financing
In the LTV analysis, it is also necessary to consider automobile financing.
Like most car companies, Asian runs an automobile leasing and finance
business as well. This is a profitable business for Asian. If we are going
to increase the repurchase rate of Asian automobiles, we will also automatically increase the profits from leases and financing. To see how this
is calculated, let’s look at a baseline automobile leasing lifetime value
table (see Table 4-8).
There are some interesting new rows on this table. Asian does not
get to finance every Asian car that is sold, of course. The financing
really depends on the market at the particular dealership. There are
71
72
Table 4-8 Car Leasing LTV
Revenue
Years 1–3
Years 4–6
Years 7–9
Repurchase rate
Refinance rate
Retention rate
Customers
Amount financed by lease
Total amount financed
Income before taxes
Total income before taxes
45.20%
62.00%
28.00%
100,000
$30,000
$3,000,000,000
$1,452.22
$145,222,000
47.90%
64.00%
30.70%
28,000
$32,000
$896,000,000
$1,582.92
$44,321,754
49.90%
66.00%
32.90%
8,596
$34,000
$292,264,000
$1,725.38
$14,831,389
$
$
$
$
$
$
Costs
Acquisition cost
Marketing costs ($12)
$
$
$
Profit
Discount rate
NPV of profit
Lifetime value
$138,022,000
1
$138,022,000
$138,022,000
$1,380.22
6,000,000
1,200,000
7,200,000
1,680,000
336,000
2,016,000
$ 42,305,754
1.33
$ 31,808,838
$169,830,838
$1,698.31
515,760
103,152
618,912
$ 14,212,477
1.64
$ 8,666,144
$178,496,982
$1,784.97
hundreds of banks and finance companies in the auto leasing and loan
business. Asian will get the financing only if it has a good offer and the
dealer pushes the sale. Asian has determined that it gets the financing
in only 62 percent of the new car purchases. For that reason, a repurchase
rate of 45.2 percent of cars translates into a 28 percent retention rate
for leases or loans.
To get the dealers to insist on Asian financing, Asian has to make it
worth their while. It does this by paying the dealer a premium of $60
per lease or loan as a commission. That is shown here as the lease acquisition cost. In addition, Asian contacts its customers with a welcome kit
and other communications to ensure that customers appreciate the value
of Asian financing and will use Asian again when the time comes. Asian
spends $12 per customer every 3 years on communications to improve
the customer retention rate.
The discount rate for automobile financing is different from that for
automobile sales. With automobile sales, the discount rate is based on
a business-to-business sale of cars to dealers, in which the manufacturer
sometimes has to wait for the dealer to pay. With automobile financing,
the discount rate is based on business-to-consumer financing. The
payments by the consumers are immediate (unless some promotion is
involved), so the waiting period is less, but the risk factor is slightly higher.
The final result is that each new customer who buys an Asian automobile and finances it with Asian returns about $1785 to Asian over 9
years in the form of profit from financing. What will happen to that
financing profit if Asian does boost the repurchase rate of B Model cars
from 45.2 percent to 51.2 percent (adding the growth from sales during
the 3-year period to the growth to 50 percent)? The results are shown
in Table 4-9. The net increase in long-term profit to Asian is almost $7
million over 9 years (see Table 4-10). And Asian did not have to do anything unusual to get this increase.
Adding the profits from the sales of new cars to the profits from the
financing, the total profit from the relationship-building program
Table 4-9 Improved Lease LTV
Revenue
Repurchase rate
Refinance rate
Retention rate
Customers
Amount financed by lease
Total amount financed
Income before taxes
Total income before taxes
Years 1–3
Years 4–6
Years 7–9
51.20%
62.00%
31.74%
100,000
$30,000
$3,000,000,000
$1,452.22
$145,222,000
54.20%
64.00%
34.69%
31,744
$32,000
$1,015,808,000
$1,582.92
$50,248,206
57.20%
66.00%
37.75%
11,011
$34,000
$374,386,196
$1,725.38
$18,998,807
Costs
Acquisition cost
Marketing costs ($12)
$
$
$
Profit
Discount rate
NPV of profit
Lifetime value
$138,022,000
1
$138,022,000
$138,022,000
$1,380.22
6,000,000
1,200,000
7,200,000
$
$
$
1,904,640
380,928
2,285,568
$
$
$
660,682
132,136
792,818
$ 47,962,638
1.33
$ 36,062,134
$174,084,134
$1,740.84
$ 18,205,989
1.64
$ 11,101,213
$185,185,347
$1,851.85
Table 4-10 Lease LTV Gain
Improvement in LTV
Years 1–3
Years 4–6
Years 7–9
New LTV
Previous LTV
Difference
$1,380.22
$1,380.22
$0.00
$0.00
$1,740.84
$1,698.31
$42.53
$4,253,296
$1,851.85
$1,784.97
$66.88
$6,688,364
73
74
should come to approximately $80 million ($73 million plus $7 million)
over 9 years. This is the way lifetime value is used to determine the
value of relationship-building programs.
That is not the end of the lifetime value story for Asian, however.
Used-Car Owners Buy New Cars
The repurchase data developed for Asian produced some unanticipated
results concerning Asian used-car owners. Many car companies do not
consider the LTV of used-car owners in their marketing programs. The
conventional wisdom concerning used-car owners is that they do not
represent a value to the manufacturer. After all, the manufacturer made
its profit when the car was originally sold to the dealer. When the car
returns to the dealer as a trade-in, the dealer makes the profit on the resale
of the vehicle. When you look at the performance of owners of used
Asian cars, however, you find something very interesting (see Table 4-11).
This table shows just one year. As you can see, 6275 of the owners
of used Asian cars in 1997 bought new Asian cars by 2002. Similar
charts were developed for each year. Putting them together, the final
result was that in the average year, 6400 owners of Asian used cars purchased a brand new Asian. The owners of Asian used cars were, in fact,
the greatest single prospect group for the purchase of new vehicles
other than new Asian car owners.
Lifetime value is forward-looking. It does not deal with the past. We
are asking, “How much profit will you make from these people over a
Table 4-11 New-Car Purchases by Used-Car Owners
Used Asian Owned in 1997
Owned in 1997
Asian bought
by 2002
New A Model
New B Model
New C Model
New D Model
New E Model
Totals
63,761
A Model
32,777
B Model
3,691
C Model
16,065
D Model
10,545
E Model
684
6,275
2,467
801
1,642
1,094
475
2,558
1,670
242
583
132
147
828
173
295
73
53
41
1,759
372
149
692
166
147
1,111
236
107
265
705
120
19
16
9
29
39
19
Table 4-12 LTV of Used-Car Owners
Years 1–3
Years 4–6
Years 7–9
51.20%
6400
$4000
$25,600,000
54.20%
3277
$4000
$13,107,200
57.20%
1776
$4000
$7,104,102
Costs
Acquisition cost
$200
$1,280,000
$ 320,000
$1,600,000
$200
$655,360
$163,840
$819,200
$200
$355,205
$ 88,801
$444,006
Profit
Discount rate
NPV of profit
Lifetime value
$24,000,000
1.11
$21,621,622
$21,621,622
$3378.38
$12,288,000
1.31
$ 9,380,153
$31,001,774
$4844.03
$ 6,660,096
1.54
$ 4,324,738
$35,326,512
$8831.63
Revenue
Repurchase rate
Customers
Net profit per car
Net revenue
period of time?” When you ask the question this way, it is clear that usedcar owners have a lifetime value for Asian. We calculated the number
of Asian cars that typical Asian car owners were buying each year. By
knowing the profit to Asian from the cars they were buying, it was possible to calculate the lifetime value to Asian of owners of used Asian
cars, which is shown in Table 4-12.
This shows that the owner of a used Asian car has a lifetime value to
Asian of $8832 over 9 years. What can you do with this information? You
can use it to calculate the value to Asian of programs designed to increase
purchases of used Asian vehicles by members of the general public. A program that was previously considered to be something that would help the
Asian dealers is actually profitable for the manufacturer and probably
deserves more effort than has previously been expended on it.
Use of Lifetime Value Data
Lifetime value has been adopted by companies of every type throughout
the developed world as a means of valuing their customers and directing
their marketing programs. It is not that difficult to compute if you have a
75
76
database that contains transaction history, and if you can determine the
revenue and costs involved in those transactions. The real value of lifetime
value is not in the numbers, but in the use that is made of the numbers.
Without imaginative strategies, the LTV numbers are useless.
Throughout this book, we will be using lifetime value tables to evaluate the effectiveness of various relationship-building strategies. Most lifetime value tables cover only 3 or 4 years. It is only in the automobile
and similar industries (farm vehicles, trucks, capital equipment), where
there is a lease or loan arrangement that typically lasts 3 or 4 years, that
we have 9-year tables like those shown in this chapter.
Once lifetime value has been calculated, it can be put into every customer’s database record. The numbers will be different for each individual. Here is how they are calculated.
Owners of 2002 B Model cars will have one value. Owners of 2003
B Model cars will have another value. This reasoning applies to the
owners of every Asian model in every year, and to every used Asian
model. It applies to Asian lease and loan holders. People who own two
or more cars have the values of these cars added together. The final
result is a unique LTV number for every customer based on everything
that customer owns.
None of this takes into account the demographics of the customers.
Is the LTV of a 50-year-old customer different from the LTV of a 30year-old customer? How would you discover that? The answer lies in
the different repurchase rates for different age groups. Instead of looking
at only the LTV by model, we can look at it both by model and by age
group. We can look at the LTV of women versus the LTV of men. We
can look at the LTV of people living in the Northeast versus those living on the West Coast. We can look at the differing LTVs of white,
black, and Hispanic owners. Where do we stop in our calculations?
We have to consider two things:
1. Do the demographics result in significant differences?
2. If so, can we use these differences to craft different relationshipbuilding strategies that will improve the repurchase rate?
The first question can be answered in short order by an analysis of the
database going back 6 years. We can tell pretty rapidly whether older
people have a higher repurchase rate than younger people, whether there
are differences between whites and blacks, whether there are differences
between East and West, and so on. In most cases, there will be only marginal differences, so we forget about those factors. You will have to look
at the numbers to decide whether you can create a marketing program
that recognizes and uses the differences.
For each possible program, we have to decide whether the cost of the
program will be sufficiently low and the benefits sufficiently high for
the resulting LTV to actually go up when the programs are put into
place. In many cases there is little we can do. There are more possible
database marketing programs that drive LTV down than there are that
drive LTV up. That is part of the reason why CRM has been such a
failure in so many industries.
Creating LTV Segments
Once we have LTV in everyone’s database record, we can begin to think
about segmentation. We have already segmented customers by model
and year, and we may also have segmented them by demographics. One
important segmentation method is by total value to the company. We
can arrange all customers by LTV from top to bottom and create segments that we call Gold, Silver, Bronze, Steel, and Lead. Our Gold customers are the most valuable ones. They have the highest lifetime value.
They probably own more of the company’s products, and they have a
higher spending rate and retention (or repurchase) rate. They are cheaper
to serve, and they tend to buy higher-priced options. They are less pricesensitive than other segments. We really want to retain these Gold customers. We must devise programs to keep them coming back for more.
We will add their segment designation to their database record so that we
can recall them easily when we want to communicate with them.
Multiple Product Ownership
Most companies have more than one product that they can sell to their
customers. Statistics from a variety of industries, such as banks, insurance, and travel, show that the retention rate is often a function of the
77
78
number of products owned. The more of a bank’s products you own,
for example, the less likely you are to drop the bank’s credit card when
you receive a competing offer. Because of the importance of multiple
product ownership, many companies have put major resources into selling their customers a second product—not only for the profit they make
from the second product, but also because the second product shores
up the retention rate for the first product. Since multiple product ownership is important, when you create a LTV table, you can segment customers by the number of products owned and create separate retention
rates, and hence LTVs, for owners of multiple products. The chart that
a bank developed is shown in Figure 4-1.
Potential Lifetime Value
As stated earlier, lifetime value is a forward-looking concept. We don’t
care about the past. What we want to know is, how much profit will
we make from this customer in the future? The number of years is
important, of course. If you knew that Arthur Hughes would be a customer for the next 40 years, whereas Bill Anderson would be a customer
for only 5 years, you might treat Arthur Hughes and Bill Anderson differently, and perhaps treat Arthur Hughes better. But you cannot know
that. Neither Arthur nor Bill knows it either. What you can know is that
Arthur Hughes is a member of a segment that has a retention rate of
60 percent. You know his lifetime value after 3 or 4 years based on his
Figure 4-1
Gain in Retention Rate
100%
retention rate
80%
60%
40%
20%
0%
1
2
3
4
5
products owned
6
7
current rate of spending, the number of products that he owns, and the
retention rate of his segment.
There is something else that you can predict, however. You can look
at all the additional products that you can sell to Arthur Hughes and
estimate the increase in his lifetime value that would occur if you were
to sell him each of several possible products. You can also predict the
likelihood of his buying each of these products, based on the response
rate of members of his segment when they are offered the products. You
can draw up a table for each customer like that in Table 4-13.
The most profitable product you could sell this customer would be
product B, with a potential LTV of $1200. But the product he would
be most likely to buy would be product C, with a potential LTV of
only $300. Potential value is computed by multiplying the 3-year
LTV by the probability of the customer’s buying each product.
Putting both together, it looks as if the next best product for Arthur
Hughes right now is product C. If you can sell him that, then go for
product E, which has the next highest total value to your company.
Banks do this analysis today. The next best product is put into every
customer’s database record. Bank employees are asked to look at these
records when they are communicating with customers so that they can
discuss the next best product. This is using a lifetime value table to
increase bank profits.
Prospect LTV
Thus far, we have discussed the lifetime value of customers only. We
have not discussed prospects. Of course, prospects have a potential lifetime value. How to go about determining this is covered in Chapter 13,
“Finding Loyal Customers.”
Table 4-13 Next Best Product
Possible
Product
Probability
of Buying
3-Year LTV
Potential
Value
A
B
C
D
E
45%
5%
85%
35%
20%
$ 400
$1200
$ 300
$ 200
$1000
$180
$ 60
$255
$ 70
$200
79
80
Using LTV to Sell Your Marketing Program
One of the most powerful uses of LTV is internal to the company. Every
marketing organization in every corporation has to come up before a
flinty-eyed chief financial officer every year to get a budget for its next
year’s activities. Before LTV analysis, this was not a very easy sell.
Advertising can point to the colorful TV ads that it produces. Sales can
point to the number of customers that were signed up last year and the
number that are projected to be signed up in the year to come. What
can marketing point to?
Here is where LTV comes into its own. In the Asian automobile case,
the database marketing manager can come to the CFO and say, “I can
bring in $56 million more profits during the next 9 years if I can have
a budget for a welcome kit and a series of strategic customer communications.” When asked to prove this statement, the marketer can trot
out the projections shown in Table 4-7 and back them up with detailed
tables like the ones shown in this chapter. Assuming that the presentation
goes well, the CFO will retain these charts and refer to them in future
budget sessions. She is likely to ask, “What was your actual repurchase
rate? What were your actual marketing costs?”
The great thing about LTV is that these numbers are all actual projections that can be audited by PricewaterhouseCoopers or Deloitte &
Touche. They are not smoke and mirrors, but projections of real events
that will actually occur and can be verified. Database marketing, therefore, becomes a combination of science and art. The art, of course, is
thinking up the strategies that will produce the improved customer
relationships, and hence the improved repurchase rate. The science is
putting all these projections together in an LTV chart that makes sense
and proves the value of your efforts.
What Works
• To compute lifetime value, you need a customer database that
includes transactions. You need to be able to estimate the revenue
and the costs in order to compute the profit from each transaction.
• You use past history to determine the retention (or repurchase) rate.
• Lifetime value can at first be computed for all customers as a
group, but once it has been developed, it should be determined for
customer segments.
• The LTV of an individual customer is derived from the LTV of
the segment that includes that customer, adjusted for the products
the customer owns and other factors that seem to determine the
customer’s particular LTV.
• Lifetime value produces numbers that can eventually be audited
after the fact. A year later, you can go back and see what the
retention (repurchase) rate was; what the spending rate was; what
the marketing expenditures, the profits, and the discount rate
were. It is not smoke and mirrors.
What Doesn’t Work
• Computing LTV using complex formulas that your top management
does not understand. If your management doesn’t understand exactly
how you developed it, it will have no confidence in it and will not
use it profitably.
• Looking at LTV numbers by themselves. An LTV number by
itself is meaningless. Its most important use is in calculating
whether proposed marketing programs will increase or decrease
lifetime value.
• Spending resources on getting an exact lifetime value number.
LTV is a future number. The future has so many unknown factors
that you should use LTV as a rough guide: Do this, and don’t do
that. That is all it is good for.
Quiz
1. Compute the discount rate in the third year. Interest rate ⫽ 6
percent. Risk factor ⫽ 1.4.
2. Use this number to compute the net present value of profits in the
third year, if the profits are $45,447,221.
81
82
3. Compute the retention rate for years 1 and 2 in the following
table, given the number of customers shown.
Customers
Retention rate
Year 1
Year 2
Year 3
312,886
219,445
188,994
4. If a company has a retention rate in the second year of 60 percent,
what will selling an additional product to the customer in that year
do to the retention rate in the third year compared to what it
would have been?
a. Make it go up.
b. Make it go down.
c. Not enough information is given to know one way or the
other.
d. It depends on the discount rate.
e. It depends on the lifetime value.
5. Which is more important in determining potential lifetime value?
a. The probability of the customer’s buying an additional
product
b. The profitability of the customer’s buying that product
c. Current lifetime value
d. Both a and b
e. None, since potential lifetime value cannot be predicted
6. What is the best way for marketing to justify its annual budget?
a. Movement in the discount rate
b. Movement in the lifetime value in future years
c. Movement in the potential lifetime value of prospects
d. Reduction in the retention rate
7. What is the next best product for a customer?
a. The one with the most margin for the seller.
b. The one with the highest probability of a sale.
c. This cannot be known in advance.
d. The one with the highest value to the company, which is
determined by multiplying the probability of purchase and
the profitability if purchased.
e. The one with the highest lifetime value if purchased.
8. If analysis shows that LTV will go down in future years, what
should you do?
a. Revise your marketing program.
b. Redo your LTV tables, since future LTV always goes up.
c. Revise the discount rate.
d. Add more marketing dollars.
e. Reduce the retention rate.
9. Suppose you have two segments. Segment A has an LTV of $150
in 3 years, and segment B has an LTV of $200 in 3 years. What
should your strategy be?
a. Work to retain segment B and increase the LTV of
segment A.
b. Get rid of segment A and find more people that fit into
segment B.
c. You can’t decide based on what we know at present.
d. Retain both, since the LTV is positive.
83
C
H A P T E R
5
The Value of a Name
T
he names and addresses of ultimate consumers have always been
valuable in direct marketing, if you know how to use them. But
not all names are valuable.
In the last 20 years, we have discovered that the names and addresses
of purchasers of packaged goods sold in retail stores have almost no
value. This is because you can’t profitably use direct mail to get people
to buy packaged goods in retail stores.
What could I do with a clean, household database of all the families
in the United States that use Ivory Soap? I could lose a lot of money.
Database marketing is incremental. You are already selling Ivory Soap
in retail stores. Your assumption is that, by sending direct-mail messages, you might sell a little bit more Ivory or a little bit more of another
P&G product, perhaps 10 percent more. When you look at the cost of
direct mail, you will soon learn that you are dead wrong. Ivory Soap is
selling in some supermarkets at $1.29 for four bars. The profit on those
four bars is obviously tiny. A 10 percent increase would be even tinier.
A good response to direct mail is 2 percent. There is no way that direct
mail could possibly increase sales sufficiently to pay for itself when promoting most packaged goods.
This is true not only for Ivory Soap but for almost everything sold
in a supermarket, office supplies store, hardware store, or department
store. Manufacturer after manufacturer has tried promotions with
rebates by direct mail. They have collected thousands of consumer
names and addresses in this way. None of them has figured out a way
to make money from those names.
84
T h e Va l u e o f a N a m e
That does not mean that consumer names are worthless for direct
mail. Catalogs obviously work extremely well. Supermarkets profitably
send circulars that get customers to come back, or to shop on Tuesday
night. Names of parents of new babies are very valuable indeed. What
does not work well is the two-step direct-mail process for regular packaged
goods: Send consumers a direct-mail letter inviting them to go to a
retail store to buy the product. If you use TV, radio, or print to do this,
it works like gangbusters.
Why are the names of packaged goods purchasers worthless? There
are several reasons:
• The margins on these products are low.
• Certificate redemption is very slow—it takes up to 6 months to
find out what is happening.
• To make a profit, you need repeat sales (low margin). You can’t
redeem a certificate for every sale, or you would go broke. So you
never find out about the subsequent sales.
• There is no way to figure out if you are succeeding. When people
buy packaged goods at a retail store, their purchase may be
captured by the store when the product is scanned. If the store has a
proprietary card, it may know that Helena Hughes bought a tube of
Crest toothpaste on August 17. But the store won’t tell Crest about
it because of privacy considerations. The small amount that Crest
might pay for the information would not compensate for the bad
public relations if Helena found out that her name was being used
and complained about it publicly. Retail stores simply will not sell
this information. They have been burned, and they know better.
• Crest doesn’t want the information, anyway, because it can’t figure
out how to use it to make a profit.
The Phone Call
A year after the dot com crash, I had a telephone call from the CEO of
a dot com start-up (yes, they were still starting them up, despite the
carnage). He told me that his new venture would generate thousands of
satisfied Web consumers who would gladly provide permission for resale
85
86
of their contact information. He asked me what the value of these names
would be, so that he could plug this number into the business plan
required by his investors. He called me because he had been told that I
had written several books and articles on the subject and could probably
provide the answers he sought, over the phone.
The question had me stammering. I had never thought about the
value of a name in the abstract like this. There are many things that
have to be pinned down before the value can be determined. In the first
place, the questions are, “Of value to whom? What are these names,
and who would want to buy them? What can these names do for a
potential buyer?” The CEO did not have the answers to these questions,
but he assumed that I did.
So I started to examine the question in detail. Let’s visualize an ideal
situation. Suppose you had the names of 500 Boeing executives earning $100,000 a year or more who have to relocate from Seattle to Chicago within the next 60 days. What is the value of these names? To a
large real estate chain in Chicago, these names could be very valuable
indeed, as Table 5-1 shows.
Assuming that the Chicago real estate chain can sell homes to 5 percent of these executives (a not unreasonable assumption), then each of
the 500 names is worth $900. The Chicago chain might be induced to
offer as much as $450 per name and could still make $450.
But this is assuming that the names can be sold only to a real estate
dealer. The executives will be buying many new services in Chicago
after they move, including
• Moving services
• Doctor and dentist
Table 5-1 Chicago Home Buyers
Amounts
Names
Home value
Commission rate
Commission revenue
Success rate
Net profit
Name value
Name offering price
500
$300,000
6%
$9,000,000
5%
$450,000
$900
$450
• Bank and insurance services
• Automobiles and furniture
These names are valuable to many different businesses. Their value
certainly exceeds $450 when all these potential buyers are factored in.
Clever and aggressive marketing could yield a substantial profit from
selling the names.
Is such a list of names of executives who are moving unique to the
Boeing situation? Not at all. Companies shift their executives around
all the time. But my dot com CEO’s list of names was not a list of highincome executives. It was a list of young people between the ages of 15
and 35 who had purchased music over the Web. What would be the
value of these names? This is a different proposition altogether. We have
to ask ourselves, “Who will want to buy these names?” The Chicago
buyers that we mentioned before would probably not be interested. Perhaps the names could be sold to other e-commerce sites. Amazon or
CDNow might buy them. Some retailer or cataloger might want them.
Here the value would be much lower. Let’s see how the marketer could
go about calculating the value of a rented name.
We can begin with the lifetime value of the acquired customers.
Table 5-2 computes the lifetime value to a retailer selling music CDs
where the average sale is $20.00.
Table 5-2 CD Buyers
Year 1
Year 2
Year 3
Retention rate
Customers
Average sale
Sales per year
Revenue
40%
200,000
$20.00
3
$12,000,000
50%
80,000
$30.00
4
$9,600,000
60%
40,000
$40.00
5
$8,000,000
Cost %
Costs
Marketing ($10/yr)
Total cost
80%
$ 9,600,000
$ 2,000,000
$11,600,000
75%
$7,200,000
$ 800,000
$8,000,000
70%
$5,600,000
$ 400,000
$6,000,000
$400,000
1
$400,000
$400,000
$2.00
$1,600,000
1.08
$1,485,608
$1,885,608
$9.43
$2,000,000
1.16
$1,724,243
$3,609,852
$18.05
Profit
Discount rate
NPV of profit
Lifetime value
87
88
What Table 5-2 tells us is that the average newly acquired customer
of this company is worth $18 in the third year. The details of this chart
are explained in Chapter 4, “Computing Lifetime Value.” For now, let’s
just concentrate on what it tells us. The retention rate of customers
tends to go up after the first year, as does the average order size. Of the
200,000 customers who might make a purchase, the average person
might buy three times a year. Forty percent of these purchasers would
still be customers in Year 2, making four larger purchases per year. Fifty
percent of those would last to a third year, buying even more.
The costs of servicing customers typically tend to go down each
year, as shown here. The discount rate is needed to convert future
profits into net present value of profit so that these profits can be added
together to get the cumulative profit. The lifetime value is determined
by dividing the cumulative NPV of profit by the original 200,000 customers to determine the value of a newly acquired customer. In this
case, the newly acquired customers are worth $18 in the third year.
This is the value to a company selling CDs that has already acquired
these customers.
So to the dot com the names are worth $18 to their owner. What would
they be worth to a potential buyer? Let us assume that the potential buyer
is selling a product of similar value, so that the buyer’s customer lifetime
value once it has acquired the customers would be identical—$18. What
is the value of these names to a new company that wants to sell products
to these customers? The value of a name has to be related to the cost of
acquiring the customer by using various media (see Table 5-3).
An outside buyer renting the name from the original company has
to consider the costs of acquiring these customers on its own. There are
two methods the buyer could use: direct mail or email. Let’s consider
both possibilities; the costs are given in Table 5-4.
Table 5-3 Sales by Medium
TV
Radio
Print
Direct mail
Budget
Reached
CPM
Sales
rate
Sales
Cost per
sale
$400,000
$400,000
$400,000
$400,000
74,074,074
71,428,571
25,000,000
1,086,957
$5.40
$5.60
$16.00
$368.00
0.04%
0.03%
0.07%
1.80%
29,630
21,429
17,500
19,565
$13.50
$18.67
$22.86
$20.44
Table 5-4 Two Methods of Contact
Names
Cost of message
Sending cost
Response rate
Responses
Customer lifetime value
Value of customers acquired
Value after sending cost
Prospect lifetime value
Offering price
Direct mail
Email
200,000
$0.37
$74,000
1.8%
3,600
$18.05
$64,977
($9,023)
($0.05)
0
200,000
$0.04
$8,000
1.0%
2,000
$18.05
$36,099
$28,099
$0.14
$0.07
If the response rate for direct mail is 1.8 percent, any direct-mail
offer would be a loser. This is proof of the assertion that database marketing does not work with packaged goods. But with email, profitable
database marketing becomes a possibility. The response rate will be
lower, at 1 percent, but the marketing costs are far lower. The value of
a prospect’s name is 14 cents. A marketer might be willing to offer half
that for a name.
The lifetime value sets an upper boundary for the offering price. Anyone who offers to pay the full lifetime value for a prospect’s name will
never make a profit. As a rule of thumb, we can assume that buyers
will never offer more than half the projected lifetime value to acquire a
name. So the offering price of a name with a projected LTV of $0.14
could not be higher than $0.07.
So, if the offering price of a prospect name to an e-commerce consumer name buyer is $0.07, what is the name worth to the seller? Here
we have to consider marketing costs. Buyers do not materialize from
nowhere once names are acquired. To sell names, most companies
need the services of list managers and brokers who must advertise and
get fees. The list owner will be lucky to get 75 percent of the offering
price from any one buyer. With luck, the names can be sold several
times per year. So we end up with the value of a name looking something like Table 5-5.
So we have an answer for the CEO of the new dot com. His names
are worth $0.16 each if he finds a good list manager and promotes the
names, and if everything goes well.
89
90
Table 5-5 Value of a Name
Amounts
Offering price
Marketing costs
Net revenue
Sales per year
Value of a name
$0.07
25%
$0.053
3
$0.158
As you can see, this type of analysis is not difficult to do. It should
be done before any business is created. If it had been done by the thousands of dot coms that were set up during the boom, we might not have
had the dot com wipeout and the stock market crash. Let’s hope we see
some sharp-pencil accounting in the future when “cool ideas” are
floated to venture capitalists.
Using LTV to Compute a Name Value
The lifetime value chart shown in Table 5-2 is typical of standard lifetime value charts. The details of such charts are spelled out in Chapter
4. A lifetime value chart has many uses. It can be used to determine
•
•
•
•
•
•
•
How much to spend on acquisition of customers
How much to spend on retention programs
How much to spend on referral programs
The payoff from particular marketing strategies
The rental value of a name
The value of a company, based on its acquired customer base
Which groups of customers you should try to retain, acquire, or
discard
• And many other things
Let’s try to determine the value of a name in a business-to-business
setting. Assume we have a company that sells chemicals to industrial
users. Let’s call our company WardChem. We have about 45,000 customers who spend from $5000 to $500,000 per year on chemicals. The
average order is $2500, and the average customer buys from us eight
times a year. Later we will break our customers down into segments
(low-volume, medium-volume, high-volume, etc.), but for now, we will
lump them all together into a single lifetime value table (Table 5-6), so
that we can understand the principles involved.
We are assuming newly acquired customers in year 1 and examining
their performance over the next 3 years.
• The retention rate is the percentage of customers that remain from
year to year. As we can see, WardChem loses 30 percent of its newly
acquired customers after the first year. It does better in subsequent
years, with the retention rate rising to 80 percent in the third year.
• Retained customers are those same customers in their first,
second, and third years with WardChem. New customers who
arrive in year 2 or year 3 are not shown here. The year they arrive
is their year 1. Let us assume that WardChem retains a base of
about 45,000 customers. It does this by losing 9000 customers and
gaining 9000 new ones every year. We are studying the history of
the retained customers.
• The average order size tends to grow every year. This is typical of
all corporations.
• The number of orders per year also tends to grow.
• The cost percent is a rough number that estimates the average
Table 5-6 Lifetime Value for Wardchem
Year 1
Year 2
Year 3
Retention rate
Retained customers
Average order
Orders per year
Total revenue
70%
45,000
$2,500
8
$900,000,000
75%
31,500
$2,700
9
$765,450,000
80%
23,625
$2,900
10
$685,125,000
Cost rate
Variable costs
Acquisition cost ($1,400)
Retention costs ($90/yr)
Total costs
75%
$675,000,000
$ 63,000,000
$ 4,050,000
$742,050,000
71%
$543,469,500
0
$ 2,835,000
$546,304,500
69%
$472,736,250
0
$ 2,126,250
$474,862,500
Gross profit
Discount rate
Net present value of profit
Lifetime value
$157,950,000
1.01
$155,779,959
$155,779,959
$3,462
$219,145,500
1.09
$200,894,344
$356,674,303
$7,926
$210,262,500
1.17
$179,159,636
$535,833,940
$11,907
91
92
percentage of the selling price of company products that goes for
manufacturing, distribution, and other costs. The balance is profit
or marketing costs.
• The acquisition cost is computed by adding together all the
expenses involved in acquiring new customers during a given
year—advertising on TV, radio, print, and direct mail; sales
force salaries, commissions, and bonuses—and dividing that
total by the number of customers actually acquired during that
year.
• The retention costs are those costs spent to keep customers
happy and buying. This could include an annual retreat,
customer communications, or customer services.
• The discount rate is used to compute the net present value of the
profits received. Money that will arrive in 1 year or 2 years is
obviously not as valuable today as money that arrives today. The
discount rate is a number that puts all these amounts on the same
basis so that they can be added together. As discussed in earlier
chapters, the formula for the discount rate is
D ⫽ (1 ⫹ i * rf )waiting time
In this formula,
i is the interest rate that the company pays for money
rf is the risk factor (the risk that the money will not materialize
at all)
• The waiting time is the length of time you have to wait before the
money arrives. With consumers, this is usually 0 for the first year,
1 year for the second year, and so on. With business-to-business
selling, the picture is different. Most business-to-business
customers pay in 30, 60, or 90 days. So the waiting time is
determined by adding these periods to the years.
Table 5-7 is a simple chart that is used for calculating the
discount rate of a business-to-business situation.
• The net present value of profits is the gross profits divided by the
discount rate.
• The cumulative NPV of profits is calculated by adding the profits
from previous years to this year’s profits.
Table 5-7 Discount Rate Calculation
Years to wait
Interest rate
Risk factor
Accounts receivable days
Discount rate
Year 1
Year 2
Year 3
0
6.80%
1.1
70
1.01
1
6.80%
1.1
75
1.09
2
6.80%
1.1
80
1.17
• The customer lifetime value is calculated by dividing the cumulative
NPV of profits in each year by the total number of newly acquired
customers—in this case, 45,000. In this example, newly acquired
customers are worth $11,907 in their third year with WardChem.
Calculating LTV by Segments
Calculating LTV for customers as a whole is useful. Calculating the
LTV of customer segments is even more useful. An automobile company may find that the LTV of various models may be quite different,
as shown in Chapter 4. Two factors, in general, are responsible for the
difference:
1. The manufacturer’s profit per car
2. The repurchase rate
With that knowledge, it is possible to determine the additional total
profit that a company would earn if the repurchase rate could be modified by changing the behavior of customers. This knowledge can lead
to the development of marketing programs aimed at particular customer segments.
Increasing the Retention Rate
All the factors in a lifetime value table can be influenced by marketing
programs: the average order size, the number of orders per year, and
the retention rate. To see the way this works, let’s focus on only one
93
94
factor, the retention rate, and see what would happen if WardChem
could increase that rate by 5 percent—an achievable goal for most companies. Let’s see how WardChem could go about it.
•
•
•
•
It could set up a relationship-building team for its top customers.
It could provide a much higher level of customer service.
It could set up a super-customized Web page for each customer.
It could put thousands of pages of technical information about its
chemicals on the Web for its customers (only) to use.
Let’s assume that the cost of all these additional relationship-building
activities is about $110 per customer per year, or about $5 million.
There could be two results: The retention rate might go up by 5 percent,
and the number of orders might go up by one-half order per year. Table
5-8 shows what could happen.
Lifetime value in the third year has gone up from $11,907 to
$13,362. What does that mean for WardChem in terms of profits? As
Table 5-9 shows, it means more than $65 million more profits—after
taking into account the extra $5 million per year spent on relationshipbuilding activities.
Table 5-8 Revised Lifetime Value
Year 1
Year 2
Year 3
Retention rate
Retained customers
Average order
Orders per year
Total revenue
75%
45,000
$2,500
8.5
$956,250,000
80%
33,750
$2,700
9.5
$865,687,500
85%
27,000
$2,900
10.5
$822,150,000
Cost rate
Variable costs
Acquisition cost ($1,400)
Retention costs ($200/yr)
Total costs
75%
$717,187,500
$ 63,000,000
$ 9,000,000
$789,187,500
71%
$614,638,125
0
$ 6,750,000
$621,388,125
69%
$567,283,500
0
$ 5,400,000
$572,683,500
Gross profit
Discount rate
Net present value of profit
Lifetime value
$167,062,500
1.01
$164,767,265
$164,767,265
$3,661
$244,299,375
1.09
$223,953,322
$388,720,586
$8,638
$249,466,500
1.17
$212,564,425
$601,285,011
$13,362
Table 5-9 Gains through Retention-Building Activities
New LTV
Old LTV
Difference
Year 1
$3,661
$3,462
$200
$8,987,305
Year 2
$8,638
$7,926
$712
$32,046,283
Year 3
$13,362
$11,907
$1,454
$65,451,072
Lifetime value charts like these are used in calculating the value
of a name. They are very powerful predictors of success in database
marketing.
The Value of an Email Name
A name and address with the matching email name and permission to
use it can be a very valuable commodity—much more valuable than a
simple name and address. There are many reasons why this is so:
• It is far cheaper to send email communications than direct-mail
communications, so you save money on each communication.
• Since emails are so cheap and so fast, you can undertake many
communications that you simply could not do at all with direct mail.
• Using emails, you can build close relationships that you could not
build if you were limited to direct mail or phone. As a result, the
email will increase the retention rate of certain customers, which
adds directly to their lifetime value.
• Emails can be used in viral marketing. What this means is that
you can get customers to write to one another and pass your
message to other people whom you don’t know.
• The speed of emails permits you to send types of messages that could
not be sent by any other medium. Airlines, for example, send weekly
messages about the availability of unsold seats at rock-bottom prices.
These messages could not be sent by mail because of the time issue,
or by phone because of the cost and the annoyance factor.
Let’s put all these factors together as shown in Table 5-10 to begin
to estimate the value of an email name.
95
96
Table 5-10 Annual Email Name Value
Regular promotion
Low-cost item promotion
Last-minute special
Retention messages
Follow-up messages
Viral marketing
Newsletter
Rental value
Total annual value
Direct
mail value
Email
value
$0.80
NA
NA
$2.40
$1.20
NA
$1.20
$0.42
$6.02
$1.28
$0.32
$2.48
$3.68
$9.12
$0.24
$2.24
$4.20
$23.56
These numbers are very important, as they lie at the heart of the
relationship between modern database marketing and the Web. Let’s
discuss each type of promotion in turn.
Regular Promotions
Message for message, email is cheaper to send. Victoria’s Secret has
used email marketing for several years. It has more than 4.5 million customer email names in its database. It outsources its email to Digital
Impact. The emails are timed to coincide with Victoria’s Secret catalog distribution. This type of email generates a purchase rate of about
6 percent, according to Shar Van Boskirk of Forrester Research.
Suppose you send regular promotions eight times per year to your
customer base. Your response rate might look like Table 5-11. As you
can see, the response rate for direct mail is higher than that for email.
This will normally, although not always, be true. However, the higher
cost of direct mail swamps the improved response rate, giving the profit
per name from email a 60 percent lift.
If you do eight promotions per year, your email addresses are worth
$1.28 per year, or $0.48 more than a simple name and address.
Low-Cost Item Promotion
There are many items that you simply cannot profitably promote by
direct mail. One example is the Universal Music promotion of music
Table 5-11 Profit per Name from Regular
Promotions
Sent
Cost for each
Promotion cost
Response rate
Responses
Profit per response
Profits
Net profits
Eight promotions per year
Annual profit per name
Direct mail
Email
200,000
$0.50
$100,000
3.0%
6,000
$20.00
$120,000
$20,000
$160,000
$0.80
200,000
$0.04
$8000
1.0%
2,000
$20.00
$40,000
$32,000
$256,000
$1.28
Table 5-12 Profit per Name from Low-Cost
Item Promotion
Sent
Cost for each
Promotion cost
Response rate
Responses
Profits
Net profits
Eight promotions per year
Direct mail
Email
200,000
$0.50
$100,000
3.0%
6,000
$8.00
$48,000
⫺$52,000
NA
NA
200,000
$0.04
$8000
1.0%
2,000
$8.00
$16,000
$8000
$64,000
$0.32
CDs, which sell for about $20 each. The profit to a company selling a
$20 item directly is probably on the order of $8 per item. Table 5-12
shows the way the promotion could work out.
Using direct mail for this type of promotion is a loser, so no one would
attempt it. Each email name, on the other hand, earns a healthy $0.32 per
year. This is found money that comes from having email names.
Last-Minute Special
Only certain types of companies have last-minute specials. These include
airlines, railroads, hotels, rental car companies, cruise lines, and producers of live theater, sporting events, and concerts—any company that
sells something that disappears if it is not used. Last-minute specials also
97
98
Table 5-13 Profit per Name from
Last-Minute Special
Sent
Cost for each
Promotion cost
Response rate
Responses
Profits
Net profits
Profit per name
Fifty promotions per year
Direct mail
Email
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
200,000
$0.04
$8,000
0.5%
1,000
$70.00
$70,000
$62,000
$0.31
$2.48
apply to retail stores that are having a special sale that comes to an end
in a very few days. These promotions cannot be sent by regular mail
because it is too slow. Table 5-13 shows what the value looks like.
The profit from a last-minute special is pure profit. The seats are
going empty. It costs very little to let people sit in them. Even at a very
low response rate, the profits are substantial, and the annual value of a
name is considerable.
Retention Messages
Every database marketer likes to think up retention messages. What are
retention messages? They are such things as birthday cards, anniversary
cards, season’s greetings, thank-you messages, customer surveys, service reminders, quarterly savings summaries, etc. The value of these messages is not in the response, since none is requested or expected. Their
value is derived from the fact that they keep customers from defecting
by building profitable relationships with them. The profit comes from
the increased retention rate and the consequent increased lifetime value.
To understand this, let’s set up a control group that does not get the
retention messages. Table 5-14 shows the lifetime value of this group.
This represents customers of a women’s dress chain that has several
stores. It costs the chain $170 to acquire women who make two visits
to the chain per year, spending an average of $160 per visit. The chain
retains only 40 percent of these women into the second year. Of those
retained, 50 percent remain to the third year. For the loyalists who are
Table 5-14 Control LTV for Women’s Dress Chain Customers
Year 1
Year 2
Year 3
Customers
Retention rate
Visits per year
Amount spent per visit
Revenue
200,000
40%
2
$160
$64,000,000
80,000
50%
3
$200
$48,000,000
40,000
60%
4
$220
$35,200,000
Cost rate
Cost
Acquisition cost ($170)
Direct mail marketing messages @ $0.50
Marketing costs
Total cost
60%
$38,400,000
$34,000,000
6
$ 600,000
$73,000,000
58%
$27,840,000
56%
$19,712,000
6
$ 6 00,000
$28,440,000
6
$ 600,000
$20,312,000
Profit
Discount rate
NPV of profit
Lifetime value
⫺$9,000,000
1
⫺$9,000,000
⫺$9,000,000
-$45.00
$19,560,000
1.07
$18,246,269
$9,246,269
$46.23
$14,888,000
1.15
$12,955,280
$22,201,548
$111.01
still shopping at the chain, the number of visits and spending per visit
tend to grow. The chain spends $4.00 per year on direct-mail marketing messages, including seasonal specials. It loses money on women in
the first year, but overall it earns a good profit from those who remain,
so that the lifetime value of a newly acquired customer after 3 years is
$111. The chain is doing well.
Now we will determine the effect of additional email retention messages on a similar group of women, which we will call the test group.
These women get 36 personalized email retention messages per year
from the store, in addition to the regular direct mail. These messages
cost $0.04 per message. They have the effect of increasing the retention rate by 5 percent, the number of visits per year by 0.2 visit, and the
spending rate per visit by $5. Table 5-15 shows the results.
As you can see, these retention messages have a dramatic effect on
the women’s behavior. Note that we could not have had this effect by
simply increasing the number of direct-mail messages, which were costing us about $0.50 per message. Had we tried that, 36 messages per year
would have cost us $18 per customer, or $3.6 million per year, which
would have seriously eaten into profits. In addition, many of these email
messages were quite timely and could not have been sent at all by direct
99
100
Table 5-15 Improved LTV through Retention Messages
Year 1
Year 2
Year 3
Customers
Retention rate
Visits per year
Amount spent per visit
Revenue
200,000
44%
2.2
$165
$72,600,000
88,000
54%
3.2
$210
$59,136,000
47,520
64%
4.2
$240
$47,900,160
Cost rate
Cost
Direct mail marketing messages @ $0.50
Marketing costs
Email messages @ $0.04
Email total
Total cost
60%
$43,560,000
$34,000,000
6
$ 600,000
36
$ 288,000
$78,448,000
58%
$34,298,880
56%
$26,824,090
6
600,000
36
$ 126,720
$35,025,600
6
600,000
36
$
68,429
$27,492,518
⫺$5,848,000
1.00
⫺$5,848,000
⫺$5,848,000
⫺$29.24
$24,110,400
1.07
$22,491,045
$16,643,045
$83.22
$20,407,642
1.15
$17,758,376
$34,401,421
$172.01
Profit
Discount rate
NPV of profit
Lifetime value
$
$
Table 5-16 Gains through Retention
Control
Test
Gain from use of email
With 200,000 customers
Year 1
Year 2
Year 3
−$45.00
−$29.24
$15.76
$3,152,000
$46.23
$83.22
$36.98
$7,396,776
$111.01
$172.01
$61.00
$12,199,873
mail. The effect of the email can be seen in Table 5-16. The gain to the
chain is $12 million in pure profit in the third year.
Now we can evaluate the retention messages as shown in Table 5-17.
As you can see, increases in retention and lifetime value are much more
difficult to measure than other direct marketing factors. You have to set
up control groups that do not get the retention messages and then see,
after a year or so, whether there is a difference in the retention rate and
the lifetime value. The only difference between email and direct mail
here is that direct mail is probably more effective in building retention
than email. A thank-you letter that is printed and mailed means more
than a thank-you email. The same might be true of birthday cards and
other retention messages. On the face of it, however, it is a lot easier to
get management to spring for $8000 for email birthday cards than for
$100,000 for direct-mail ones. The probability is that emails will be
much more widely used as people come to see their value. So the email
will probably be the clear winner in retention messages.
Follow-Up Messages
Here is an area where email wins in a walk. When people order something, you follow up with an email confirmation (or a mailed confirmation). When the product is about to be shipped, you email a notice
with the tracking number (it is too late to do this by mail). When the
product arrives, you send an email survey: Did it arrive OK, and was it
to your liking? Here, again, the email wins. The value is measured in
customer retention and increased lifetime value, as shown in Table 5-18.
Table 5-17 Name Value from Retention
Messages
Sent
Cost for each
Promotion cost
Increased retention
Current LTV
Increased LTV
Total LTV increase
Net profits
Eight messages per year
Annual name value
Direct mail
Email
200,000
$0.50
$100,000
0.080%
$111.01
$0.80
$160,000
$60,000
$480,000
$2.40
200,000
$0.04
$8,000
0.050%
$111.01
$0.50
$100,000
$92,000
$736,000
$3.68
Table 5-18 Name Value from Follow-Up
Sent
Cost for each
Promotion cost
Increased retention
Current LTV
Increased LTV
Total LTV increase
Net profits
4/12 per year
Annual name value
Direct mail
Email
200,000
$0.50
$100,000
0.01%
$111.01
$0.80
$160,000
$60,000
$240,000
$1.20
200,000
$0.04
$8,000
0.01%
$111.01
$0.80
$160,000
$152,000
$1,824,000
$9.12
101
102
You may be able to send only 4 direct-mail follow-up messages per
year, but you can send 12 or more emails, depending on the number of
transactions. The effect on retention and LTV from each message
would be the same to the customer.
Because of the cost of sending follow-up messages by direct mail, very
few companies send such messages. A better comparison than email
versus direct mail for follow-up messages would be to test email versus
no follow-up at all. The boost in retention can be measured over a period
of a year and can be quite substantial.
Viral Marketing
This is an interesting concept that works only in certain product situations. You can get some customers so excited about some product that
they will write to their friends about it. For what kinds of products or
situations does viral marketing work?
•
•
•
•
•
•
•
Concerts
Firearms
Automobiles
Travel destinations
Books
Political candidates
Health foods
As you can see, the list is long. How does viral marketing work?
As shown in Figure 5-1, you send an email to customers inviting
them to invite other customers by entering their names on a microsite
(shown in Figure 5-2). This site, which is connected to your database,
records the names and suggests messages to be sent to the names. When
the customer clicks OK, the messages are sent, and if the recipients of
the message click here, they see a microsite inviting them to accept the
offer, thus helping both the customer and themselves.
In this case, one email results in the capture of more names and ultimately thousands of other emails, all promoting your product. Viral
Figure 5-1
Viral Marketing
Susan: If you want to
have some friends
come to Hawaii with
you, list them here, and
if they decide to go,
we will give each of you
a 10% discount on your
trip.
Viral email results in
more emails and
data capture
Arthur Hughes ahughes@computersrategy.com
Helena Errazuris micifuz2@aol.com
Dudley Foulke dudleyf@hotmail.com
Dear Helena: Rick and I are
going to Hawaii on August
7. We thought it would be
great if you could come with
us. We are staying at the
Maui Surfside, which is a
great deal. To find out about
it, Click here! Hope you can
join us! Susan
Figure 5-2
Name
Database
Viral Email Dispatch
Email
Click Here to see Message
Send
Send
Send
Send
Send
Send
marketing is unique to email marketing. It cannot be effectively done
by direct mail or phone. The name value is shown in Table 5-19.
Viral marketing can be used for very inexpensive items or for quite
expensive ones. The cost per sale of $16.67 shown here assumes that
the viral messages sent by customers to their friends will have a higher
103
104
Table 5-19 Name Value from
Viral Marketing
Sent
Cost for each
Promotion cost
Response rate
Responses
Emails captured
Emails sent
Success rate
Sales @ $100
Sales
Cost per sale
Value of a name
Direct mail
Email
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
200,000
$0.04
$8,000
2.0%
4,000
1.5
6,000
8%
480
$48,000
$16.67
$0.24
Table 5-20 Name Value from Newsletters
Sent
Cost for each
Promotion cost
Response rate
Responses
Average sale
Profit per sale
Profit from newsletter
Net profit
Twelve per year
Direct mail
Email
200,000
$0.50
$100,000
2.00%
4,000
$100.00
$40.00
$160,000
$60,000
$240,000
$1.20
200,000
$0.04
$8000
1.50%
3,000
$100.00
$40.00
$120,000
$112,000
$448,000
$2.24
response rate than messages sent to prospects or to customers in general.
Viral marketing can be very powerful if it is artfully used.
Newsletters
Email newsletters have become very popular. They save a lot of money.
They often get a good response rate. It should be easy for recipients to
discontinue them with one click. That is not so easy for the recipients
of direct-mail newsletters. Newsletters result in increased retention,
sales, and loyalty. They are usually worth the effort if you can put something into them that your customers want to read. The name value is
shown in Table 5-20.
Name Rentals
Of course, when you have a file of opt-in names of customers who have
agreed that they will accept promotions from your business partners, there
is another valuable use of the email names you have collected. Since you
are using your names more effectively, you can assume that others will be
doing the same. Emails have caught on. Billions of commercial email messages are being sent every week. There is a massive market for opt-in
names. If you can rent your direct-mail names 8 times a year, you can certainly rent your email names 10 times as often, or 80 times per year. The
price per thousand email names varies widely and will certainly be very different from the time of writing to the time that you are reading this book.
Let’s take a conservative estimate of $70 per thousand for both direct-mail
and email names with permission to use them for partner messages. Table
5-21 shows what you can expect to receive during a year for each name.
Let’s summarize again the value of a name in Table 5-22.
Of course, few, if any, companies can use every one of these marketing methods, but your company can certainly use at least half of
Table 5-21 Value of Name Rental
Rental value
Rentals per year
CPM
Revenue
Commissions
Net revenue
Direct mail
Email
8
$70.00
$0.56
$0.14
$0.42
80
$70.00
$5.60
$1.40
$4.20
Table 5-22 Total Name Value
Regular promotion
Low-cost item promotion
Last-minute special
Retention messages
Follow-up messages
Viral marketing
Newsletter
Rental value
Total annual value
Direct
mail value
Email
value
$0.80
NA
NA
$2.40
$1.20
NA
$1.20
$0.42
$6.02
$1.28
$0.32
$2.48
$3.68
$9.12
$0.24
$2.24
$4.20
$23.56
105
106
them. We can say with some confidence that email names, with permission, are worth at least $15 to $20 each to any company. To my
knowledge, no company has done this analysis and therefore knows,
with any kind of precision, what emails are worth to it. What does that
number tell you in terms of marketing strategy? You must collect email
names and get permission to use them.
How do you do that? By every means at your disposal:
• Telemarketers must be compensated for capturing these names.
• Every form should be revised to include a space for email names.
Don’t wait for the forms to run out. Junk them and print new
ones.
• Run contests in which, in order to enter, customers must go on
the Web to a microsite and enter an email name and provide
permission to use it.
• Be creative. You gain $10 per year for every email name captured.
• Lenscrafters gives $10 for each email name provided by customers.
We are in a new millennium, folks, and the rules of the game have
changed. Like the invention of catalogs, direct mail, radio, or TV, this
new phenomenon, email, has fundamentally changed the nature of marketing, and we must change with it.
Your Action Plan
1. Now that you have read this chapter, go to the Web site
www.dbmarketing.com and download all the tables in this chapter.
(Incidentally, when you do this, you will be asked to give your
email name to the Database Marketing Institute. How’s that for
email name capture?) Using these tables, figure out how much
your company can make by using email more effectively.
Determine the value to your company of an email name.
2. Develop an action plan to capture more email names. Get a
budget for it and sell it within your company. Establish your
schedule and determine when you will get the names.
3. With the names in hand, develop programs that will use these
names effectively to build sales and your retention rate.
What Works
• Using test and control groups for all database marketing initiatives
• Determining the value of your email names
• Determining customers’ lifetime value and using it to evaluate
your marketing strategy
• Capturing customers’ email names with permission to use them
• Viral marketing
• Retention marketing
• Emailing last-minute specials
• Emailing follow-up messages
• Emailing newsletters
What Doesn’t Work
• Failing to capture customers’ email names and use them in your
marketing program
• Using database marketing for packaged goods, with some exceptions
• Paying more than half the LTV to acquire a rented name
• Starting any dot com without first computing the value of the
email names to be acquired
• Thinking that email response rates will be as high as direct-mail
response rates
• Failing to use control groups with any database marketing program
Quiz
1. Compute the lifetime value, after 3 years, of newly acquired
customers given these facts: 400,000 customers; retention rates
62 percent, 70 percent, 74 percent; average purchase $320,
107
108
2.
3.
4.
5.
6.
7.
8.
9.
10.
$340, $350; cost percent 65 percent, 63 percent, 62 percent;
acquisition cost $140; no marketing costs; market rate of interest
7 percent; risk factor 1.5.
Develop a second table to show the effect of a 4 percent increase
in the retention rate, an increase in annual sales of $12, and
marketing costs of $6 per customer per year, with everything else
the same.
Show the profit or loss to the firm after 3 years if it makes the 4
percent increase. Assume 400,000 customers.
Determine the value of these names for resale, assuming
brokers’ fees of 25 percent and assuming that the buyers use
email, with a 1.2 percent success rate. Assume four rentals of
names per year. Email costs $0.05 each, and the LTV of the
names is the same as that developed in Question 1. Buyers offer
50 percent of the value of the names to them.
List five profitable uses for email names in your own situation.
Estimate the annual value of the email names resulting from
each use.
Provide five reasons why regular packaged goods names are
worthless.
List five specific action steps to be taken in your company (or
your client’s company) to obtain email names. Estimate how
many you will capture during the next 12 months.
Come up with a viral marketing plan for your company, listing
what you will promote and providing the text of the viral
message to be sent by your customers to their friends.
Of your current junk emails, find out how to unsubscribe from
ten of them. List the percentage that was easy to do and the
ones that were hard.
Think of a last-minute special email that would work for your
company. Write the sender’s address, the email subject, and the
first five sentences of the email.
C
H A P T E R
6
The Power of
Communications
Any communication, irrespective of whether or not there is a promotional offer, will increase visits.
Judd Goldfeder, The Customer Connection
C
ustomer communications work. They get people to come back,
improve cross sales, and result in upgrades and higher retention
rates. And now, because of the Internet, their cost is at an all-time low.
To see how communications work, let’s look at a few examples.
Business-to-Business Relationship Building
A lighting manufacturer based in Milwaukee used a catalog to market its
products to 45,000 contractors and builders. Business was both good and
profitable. A consultant at Hunter Business Direct thought that sales could
be made even higher through the use of customer communications. His
suggestions were greeted with skepticism, but higher management was
persuaded to give him a 6-month test. He took the top 1200 customers
and divided them into two similar groups: a test group and a control
group. The control group was given the same wonderful service that was
given to all other customers: Wait for the phone to ring and fulfill the
order rapidly.
For the test group, the consultant set up a two-person communications team. A lighting engineer was paired with a customer service rep.
109
110
Their job was to call each one of the 600 companies in the test group,
find out who the decision makers were, and talk to them. The purpose of
the calls was not to make sales. In fact, the team was not empowered to
offer discounts. The purpose of the calls was to discuss the customers’
lighting needs. Were there new products on the market that they did
not find in the catalog? What problems had they had in installations?
Where did they see the industry going? Did they need training? Did
they have suggestions?
This went on for 6 months, after which the consultant did a complete report on the results. His findings were truly amazing. In the first
place, during the 6 months, the test group placed 12 percent more
orders than they had in the previous 6 months. The control group
placed 18 percent fewer orders!
The test group’s average order size grew by 14 percent, whereas the
control group’s order size shrank by 14 percent. In total, the test group
purchased $2.6 million more product than the control group did, spending 27 percent more than they had previously, whereas the control
group’s total purchases fell by 30 percent. The total cost of the experiment was about $60,000, including the salaries and expenses of the two
communicators and the cost of their telephone calls. The increase in
sales per dollar spent was a staggering $43.
The heart of this test was the brilliant plan of setting aside the 600company control group. Without this group, no one would have
known how successful the communications program was. This is an
experiment that can be repeated in virtually any business-to-business
marketing situation.
Consumer Relationship Building
The lighting company’s experiment would not have worked with businessto-consumer selling because consumer spending is
• Not continuous
• Far less predictable
• Much lower in volume and margins
T h e P o w e r o f C o m m u n i c at i o n s
However, consumer communications can also produce marvelous
results. A chain of restaurants based in San Diego, advised by Judd
Goldfeder at The Customer Connection, gave patrons a paper membership card, which was scanned when they arrived at one of the restaurants and used to award them points toward free meals. The program
seemed to be working well. After the chain had issued 25,000 cards, it
decided to replace the paper cards with plastic ones. When the new
cards were ready, they were mailed out with a letter thanking the patron
for his or her business. There was no offer or discount connected with
the plastic card—just a simple replacement. The results were dramatic.
In the week before the cards were sent out, these customers visited the
chain 1050 times. During the subsequent 13 weeks, this same group of
patrons visited the chain an average of 1400 times per week. Sales to
these customers increased by $12,000 per week, or $156,000 in total
over the 13 weeks. The increased sales per dollar spent were $8.32.
Keeping the Advertisers
Verizon Information Services was experiencing high attrition in its
sales of Yellow Pages advertising. The annual attrition ranged from 10
percent among long-term advertisers to 27 percent among new advertisers. The attrition amounted to over $250 million per year, which
translated to over $1.4 billion over the life of the lost advertisers. This
loss plus a lower than average acquisition rate was reducing Verizon’s
penetration in the marketplace.
A study by the Verizon Database Marketing staff showed that there
appeared to be two reasons for the high attrition rate:
1. Many small- to medium-sized businesses felt that Verizon had not
established a satisfactory relationship with them. The Verizon sales
representatives would make contact with their clients only once or
twice a year. As a result, some business owners felt that they lacked
a close working relationship with Verizon.
2. Small businesses often had difficulty understanding and calculating
the value of Yellow Pages advertising. Figuring out how to assess
their return on investment was difficult for them.
111
112
To address these issues, Verizon wanted to create a retention communications program with these objectives:
• Position Verizon as a business partner for the small-business owner,
not just another Yellow Pages provider looking to sell some ad space.
• Educate advertisers about the value of Yellow Pages.
• Target offers and messages to customers providing useful business
information.
Verizon’s challenge, like that of many other companies, was to get its
hands on good customer information in a timely manner. There was an
abundance of data on the customers in Verizon’s data warehouse, but it
was difficult to make that information available quickly enough to enable
the company to carry out time-sensitive campaigns. Some of the issues
facing Verizon included the following:
• Verizon was unable to coordinate the data with the sales cycle and
customer profile. The sales cycle is difficult because Verizon
produces more than 1500 different books during the year, with
varying closing dates. Some advertisers are in only one book, but
many are in two or more books that have different closing dates.
• It was difficult to determine the correct address for advertisers with
multiple locations.
• There was no place in the existing database to store model scores
or to segment categories.
• There was no process for communicating profile and research
information to the supporting groups and departments.
The Verizon Database Marketing staff decided to launch a new program to solve these problems. When the staff began, several solutions
were under development or already in place. Behavioral models had been
built to predict advertising purchase behavior, including models showing
• Propensity to acquire
• Propensity to cancel, decrease, or increase current spending
• Propensity to buy additional products
• Lifetime value
Some preliminary work had been done on building a data mart, using
periodic feeds from the data warehouse. This data mart would be used
to organize, summarize, and analyze customer information. Finally,
some direct-mail testing had been done, with varying levels of success.
Since there was no formalized program in place, these elements were
considered nothing more than research and development projects that
might or might not yield some long-term value.
From Zero to Hero
The breakthrough solution came as a result of cooperation between
information technology (IT) and marketing. A cross-functional team
consisting of marketing and IT professionals came together in order to
create a “Best in Class” database marketing program.
In order to launch the program, IT revised the data mart so that
it could
• Aggregate customer transaction data to an annual level to match
the sales cycle.
• Tag customer performance versus the previous year (increase,
decrease, cancel, etc.).
• Overlay business demographics (SIC code, number of employees,
annual sales, etc.).
• Attach model scores to each customer record.
• Record campaign results attached at the customer level.
When this data mart was combined with campaign management
software, it became a powerful tool for direct marketing and marketing research.
While the data mart was being built, customer models were updated
and made ready for use in segmentation and targeting. This gave the marketing team the ability to analyze and segment the customer base, design
113
114
a database marketing program that would target the segments offering
the highest opportunity, and create messaging to address those segments.
The final piece of the puzzle was to add a direct marketing agency to
the team. The selection of an agency was important in order to ensure
•
•
•
•
Strategic input
Creative insight
Tactical performance
Flawless execution
The addition of a direct-mail fulfillment specialist helped to address
these issues. Its ability to work with and provide insight to the creative
agency, as well as efficiently executing the program’s primary directmarketing campaigns, provided the foundation for success.
The End Result
The strategy was to use the database to send a series of direct-mail
pieces that would stand out in the small-business owner’s mailbox. The
idea was to make them colorful and vibrant, with bold, oversized type
on all outer envelopes. Many of the pieces were designed to help the
customers calculate their return on investment from a Yellow Pages ad.
Contact frequency and messaging were based on segments defined
by the advertisers’ tenure and their propensity scores on the various
models. A minimum spending level was set as a qualification, so that the
marketers could focus their efforts on the top 50 percent of the customer
base. All contacts were timed to match various stages of the sales cycle:
• Tenure. Advertisers were divided into multiple segments on the basis
of seniority. Newer segments received more value-reinforcement and
relationship-building messages. Segments that had been customers
longer received both relationship and up-selling communications.
• Propensity scores. Model scores were then added to the segments in
order to further target messaging. Advertisers who were at high
risk for attrition received stronger reinforcement communication
focusing on educating them on the value of a Yellow Pages ad and
how to measure the success of such an ad. Advertisers who had a
high possibility of increasing spending in certain areas were given
strong sales messages in those specific areas. Those customers with
high scores in both lifetime value and risk of defection received
messages showing them how to optimize their product mix and
encouraging trial. Those who scored low in all areas received
more relationship-building messages.
All mailings were sent first class so that any pieces with bad addresses
would be returned. The returned mail was keyed into the database. The
impact of the direct-mail communications program was tremendous.
The goal was to get a 1 percent increase in sales revenue, or $6 million
in additional Yellow Pages ad sales. Control groups that did not get the
mailings were set up to measure the results. These showed that the program generated a 3.8 percent increase in overall annual sales revenue,
with an additional $13,361,000 in incremental Yellow Pages advertising.
Comparing these results with the cost of the mailings and the database,
Verizon realized a 1681 percent return on investment.
The success of this program dramatically changed the way Verizon
communicated with its advertisers. In subsequent sales cycles, a newsletter was added to increase communication with and education of the
customer. Email retention messages were incorporated to make the program more robust. Budgets for these programs were increased in the
following year.
The marketing data mart became the cornerstone not just for database marketing, but for all loyalty and database marketing programs
companywide. It became the preferred source for information on Verizon customers.
What Has Happened to RFM?
Recency, frequency, monetary (RFM) analysis has been around for more
than 50 years. It works wonderfully in managing offers to customers.
It is worthless for prospects, however, because it is based on transactions. Companies that use RFM have saved millions of dollars by not
115
116
sending communications to those who are unlikely to respond. The
principles are described in detail in my book Strategic Database Marketing.1
As an example of the way RFM works, look at Figure 6-1, which diagrams an actual mailing to 500,000 customers coded by RFM.
The company did not break even on those customers in cells below
the 0 line; in other words, the profit from sales to them was less than
the cost of mailing to them. This type of analysis is done to a small test
sample coded by RFM. From the results of that small test sample, you
can accurately predict the behavior of 2 million customers coded in the
same way. The result was that this firm mailed only to those cells that
were profitable. It doubled its response rate. This is the power of RFM.
It still is a vital tool in direct marketing using regular mail.
The Web has changed the situation, however. Since emails are so
inexpensive, there is little point in restricting offers to existing customers in order to save money. You email to everyone. RFM, however,
is still a useful way to categorize customers, even email customers.
Do Communications Change Behavior?
The Air Miles reward program is Canada’s largest coalition loyalty program. Over 60 percent of Canadian households have enrolled in the
program for free. Under this program, members earn Air Miles reward
miles by buying goods and services from a variety of merchants. The
Air Miles for Business program allows members to earn reward miles
on their business purchases as well as their personal purchases as consumers. In the spring, the Loyalty Group, the creators and managers
of the program, offered bonus reward miles to Canadian farmers in a
direct-mail campaign urging them to buy products from such merchants as United Grain Growers, John Deere Credit, AgLine, Shell, and
Goodyear.
1
RFM software is available free for downloading at www.dbmarketing.com. During the last 5
years, an average of three companies per week have downloaded this software. It works
wonderfully for direct mail. In the standard RFM coding, there are 125 cells, from 555 (the
most responsive) to 111 (the least responsive).
117
break-even index
Figure 6-1
RFM Break-Even Chart
500
400
300
200
100
0
-100
555
455
355
255
155
RFM cell
Maria Wallin, marketing manager at the Loyalty Group, arranged
for a direct-mail piece to be sent to 50,000 western Canadian farmers
who had been active in the program in the 14 months prior to the campaign. In addition, mass media ads for the campaign were placed in six
farm publications in western Canada. The goal of the campaign was to
achieve a direct-mail response rate of 5 percent and to encourage 2 percent of those contacted to make purchases at one more farm merchant
than they had used the previous spring.
Air Miles for Business turned to OgilvyOne Worldwide in Toronto
for strategic and creative development of the mailing piece, and to
Symcor Direct Response for the laser and letter shop production. Using
the Air Miles member database, they developed 72 different combinations of letter copy and 2952 certificate sheet combinations, completely
driven by past purchase behavior. The offer of bonus reward miles was
designed to encourage increased spending during the promotional
period. There was an additional incentive for each farmer to visit one
merchant whom he or she had not visited the previous spring.
A control group of 10,000 western Canadian farmers who were
members of the program but did not receive any mailings was set aside.
The success of the campaign was measured by the overall response rate,
the number of farm merchants used, and the number of reward miles
earned by those who received the mailing versus those in the control
group who did not. The result was that 12 percent of the mail universe
responded to at least one of the offers, more than double the first objective. In addition, the number of reward miles earned by those who
received the mailing was 59 percent higher than the number earned by
the control group during the period of the promotion. The same test
111
118
had been conducted in the previous year. At that time, the mailing had
produced a 46 percent increase over the control.
The second objective was also achieved, as 6 percent of the mail universe made purchases at one farm merchant that they had not used during
the previous spring. However, while the objective was achieved, the
average number of farm merchants used was the same for the mail and
control groups. So, what can we conclude from this? You can successfully use direct mail to get people to do more of what they are already
doing. It is much harder to get them to do something different.
Email Communications
The advent of email has changed the equation. Email communications
work well. Their costs are much lower than for direct mail, but so are
the response rates.
Email customer communications have become more and more
acceptable. American Airlines sends weekly notifications of low-cost
fares. Amazon sends emails promoting new books. Lands’ End promotes
new clothing. The economics are powerful.
Direct mail is expensive. If it costs you $30.50 to get a response, as
shown in Table 6-1, you had better have something profitable to sell to
the responders. On the other hand, with emails, the response rates are
much lower, but the costs are dramatically lower.
Because email is so inexpensive, it can be abused. The cost of direct
mail, on the other hand, is high, and that is good because the expense
weeds out frivolous mailers. You know that the junk mail you receive
must be profitable, or the company would not be sending it. Even if you
are not interested in what is being offered, there must be thousands of
Table 6-1 Cost per Response: Email vs. Direct Mail
Sent
Cost each
Promotion cost
Response rate
Responses
Cost per response
Direct mail
400,000
$0.61
$244,000
2.00%
8,000
$30.50
Email
400,000
$0.04
$16,000
1.00%
4,000
$4.00
other recipients who are interested. The cost of direct mail makes it
imperative that marketers design something that will be appealing to at
least 2 percent of those to whom it is addressed. On the other hand, since
email is so inexpensive, thousands of marketers who could not afford
direct mail are cluttering the Internet with billions of unsolicited commercial communications. This poses a difficulty for legitimate marketers.
Their message gets lost in the clutter of worthless marketing messages.
Direct mail has a lot of problems that are easily solved with email. It
takes a long time, sometimes weeks, to know if your message is working
with regular direct mail. Another difficulty with direct mail is that there
is no practical method for people who do not want to receive the information to remove their names. The mailer who sent a particular piece
rented the names from someone else. He probably does not have a database. Telling the mailer to remove your name is like spitting into the
wind. The mailer might keep a database of removals and run it against
the next list that he rents, but why should he? He will make no money
by doing it, and in direct marketing, profits are everything. People don’t
do things that are unprofitable. You can contact the Direct Marketing
Association and get your name entered on its do not mail list, but many
mailers do not use this list.
A good emailer, on the other hand, makes it very easy for people to
remove their names from the list. Most of the good ones have “oneclick” removal. If you click one spot at the bottom of the email, your
name is automatically removed from the list. Anyone who does this,
however, finds it an endless job. Despite constant grooming, my email
is clogged each day with at least 20 emails that I have to delete before
I get down to the good stuff. This takes me only a couple of minutes a
day, however—certainly less time than it takes me to throw out my junk
direct mail. Despite its problems, many companies are now using email
very effectively to communicate with their customers and fans.
Promoting Music Using Emails
The Universal Music Group (UMG) is the largest music corporation
in the world, with 10,000 artists making CDs for it. For many of its
119
120
artists, UMG maintains Web sites where fans can come to see pictures of the star in action, listen to snatches of the star’s music, and
register their names. UMG has been testing the use of these registered names to promote new CD releases. One example is a new CD
release called Essence by Lucinda Williams on a new label, Lost Highway. The pilot program was designed to determine whether online
promotions would have an impact on music sales, and whether database marketing strategies and tactics would work in this industry.
This was UMG’s first attempt to marry database marketing techniques with e-marketing activities. Universal introduces hundreds of
new CDs during any given year. Lucinda Williams was not a wellknown artist, and the genre of music called “roots rock” was just
emerging as a new music style. This was not a Harry Potter type of
marketing effort. UMG’s vice president of consumer information marketing, Joe Rapolla, sponsored the promotion and worked with Beth
Clough, director of partner relations at msdbm in Los Angeles, to
run the project.
The target audience included both existing fans (customers) of Lucinda
Williams and prospects who matched the fan profile. The second most
important audience was Universal Music managers in UMG’s New
Media and Common Label Operations departments. These managers had
been attempting to prove to management that online marketing could
reduce costs and find new customers.
CDs Are Sold in Retail Stores
The Internet was widely used as a channel for music sales through many
sites, such as Amazon.com, buy.com, and CDNow.com. However, more
than 95 percent of all CDs were sold at retail stores. Most of these CDs
were promoted by radio station disc jockeys. The goal of the project,
therefore, was not to sell the new CDs over the Web. Instead, UMG
wanted to use database marketing strategies to build its email customer
base, to gather key information, and to encourage album sales. The
market was filled with competing CDs. There were very few radio stations that aired the “roots rock” artists because of the small audience
for this genre. Getting rack space for this type of CD was also difficult
because it was a new genre of music and a new label. At the time the
project began, there were no systems established to measure the success
of such a program if success were to occur. UMG management had not
given the idea much attention.
The primary objective of the campaign was to develop a successful
integrated marketing strategy that would lead to incremental CD sales
during the first week of the album’s release. The campaign included the
following elements:
• Supporting the existing UMG–Lost Highway marketing strategy
• Using online and offline marketing tactics that were trackable,
accountable, and measurable
• Proving the value of database marketing in developing fan support
and sales
• Developing a model that could be replicated for other UMG labels
and artists (see Figure 6-2)
• Using research to create the ideal customer model that would
allow Lost Highway to purchase look-alike prospects
All of this had to be done in a situation in which there was no prior
research and a very limited existing database of fans. All Joe Rapolla
had was a file of a couple of hundred Lucinda fans who had registered
2 years before.
Online Focus Groups
Beth’s first task was to develop a customer profile of Lucinda Williams
fans based on the names in the existing database. She conducted online
focus groups to get qualitative research on the audience for roots rock
(see Figure 6-3). The focus groups each contained twelve participants
and lasted for 1 hour. Each participant received a $75 cash reward
through the mail after the focus group was over. Discussion in the
groups centered on the roots rock genre, along with broader topics such
as the participant’s level of involvement with favorite artists.
121
122
Figure 6-2
Lucinda Williams Landing Page
Figure 6-3
Focus Group Page
The advantage of an online focus group is that participants can
“attend” from their home or office. UMG and msdbm managers were
able to communicate directly with the moderator during the sessions.
Unlike normal focus groups, online groups require no travel expenses
or logistic setups such as meals or the renting of a hotel room. Turnaround time was faster than with traditional focus groups.
Online Microsite
Based on this analysis, msdbm created an online microsite and database to deliver the promotion, capture the data, and report the results.
The microsite included an online survey to capture information from the
registrants (see Figure 6-4).
The second phase focused on program execution. Beth created a
model that used the profile to point to the ideal prospect. Next, she did
the research necessary to find email lists of prospects who matched the
Lucinda Williams fan profile.
123
124
Figure 6-4
Reward for Survey
Email Campaign
The subsequent email campaign consisted of a series of targeted communications, beginning with an outbound email campaign to fans and
prospects promoting the release of Lucinda Williams’s album and a
special offer for those who completed an online survey. Joint online
promotions were conducted with the UMG sister company MP3 to
expand the reach of the effort. In addition, UMG conducted guerrilla
marketing efforts collaboratively with the company M80, which used
the promotion to create excitement in the marketplace, increase sales
and airplay of the CD, and ultimately drive traffic to the online promotion site.
Data on every responder were stored in the database, and the responder was profiled to see if he or she matched the anticipated type of
listener. Responders to the special offer got a follow-up thank-you email
with another survey (see Figure 6-5) and further promotion of the
Essence album, encouraging them to purchase the CD.
Special direct links to Amazon.com’s purchase page were developed
so that visitors to the microsite who wanted to do so could make a
Figure 6-5
Survey
125
126
quick purchase. In addition, the site provided the names and addresses
of local retailers who sold music. A music sampler was embedded into
the site to give prospects an idea of the quality of the new artist and the
new genre.
When the campaign was over, responders were sent a postcampaign
follow-up email to determine which of them had purchased the CD and
why or why not. Based on these responses, Beth conducted a postcampaign analysis measuring customer satisfaction and intention to
purchase the Lucinda Williams CD.
Lucinda Williams was constantly on tour during this period. The
gradually increasing fan database was used to send targeted city-specific
emails to people who lived near places where she was holding her
concerts (see Figure 6-6).
As part of the database-driven microsite development, msdbm
developed a real-time report that was tied directly to the database (see
Figure 6-7). Through this real-time report, msdbm tracked response
information, including response by list source, publication, links, and
statistical response to the online survey. UMG and Lost Highway were
issued user names and passwords so that they could access the report at
any time. The report also featured drill-down functionality so that the
client could view response by segment, such as specific lists, publications, and so forth.
As a result of the campaign, the fan database grew 960 percent
during the 3-month promotion. This campaign proved the value of
email database marketing strategies in the music industry. UMG management was stunned by the results of this coordinated effort. It loved
the real-time reporting and the ability to have a data-driven Web site.
The overall campaign response rate (combined direct and indirect) for the
172,744 people who received emails was 15.5 percent, of which 50
percent reported that they had bought the new CD. Instead of a few
hundred registered fans, the Lucinda Williams database developed
nearly 26,718 registrants who could be used in further database marketing initiatives. UMG management, impressed with the results,
decided to use this method for promoting new artists and staying one
step ahead of its competition.
Using direct emails plus MP3, the campaign exposed Lucinda
Williams to over 3 million people. The Essence CD ranked number one
Figure 6-6
Concert Tickets Email
on Amazon.com and MP3.com for several weeks in a row during the
promotion. The campaign generated interest and excitement among
other UMG labels, which wanted to execute similar campaigns for other
artists. It became the standard by which subsequent UMG campaigns
were measured.
127
128
Figure 6-7
Online Campaign Reports
Six months later, the same technique was used for the UMG launch
of an album by roots rock artist Shelby Lynne. This campaign:
• Reached 1.4 million fans and/or prospects with direct (email)
communications plus another 1.6 million via partner and offline
efforts.
• Achieved 75,749 survey respondents, including 40,097 opt-ins,
increasing the Shelby Lynne fan database by over 600 percent in
approximately 2 months.
• Achieved an 11.2 percent response rate from Shelby Lynne
fans and a 6.8 percent response rate from the Lucinda
Williams list.
• Encouraged 20 percent of the respondents to use the viral
marketing idea by referring the promotion to others. Of those
referred, 10 percent went on to complete the survey.
Emails from Racing Fans
Emails are the most inexpensive method of communication with customers ever to come along. The cheapest form of direct mail—a
postcard—will cost you at least $0.35. Email to your own customers is
almost free. But how do you collect email names? That is difficult. Here
is how one company went about the process.
Horse racing was once the most popular sport in America, but
attendance has dropped significantly over the years. Other gambling
and sport viewing options have proliferated, hurting attendance. The
National Thoroughbred Racing Association’s challenge has been to
increase track attendance and fan involvement in horse racing. It
turned to e-Dialog of Lexington, Massachusetts, to help it use the
Internet to develop a deeper relationship with core fans and to attract
new fans to the sport.
At first, the NTR A had no opt-in email addresses. e-Dialog recommended a campaign that would create an email database: the
$1,000,000 Breeders’ Cup Challenge. The challenge was launched 8
weeks before the actual race. The contest was straightforward, but
not easy: Correctly pick the winner of each of the Breeders’ Cup
races and win $1 million. A unique hurdle for this program was that
the final roster for each of the eight races was announced just 72
hours before the big day. All contestants had to submit their eight
picks during that 72-hour window. For this reason, it was critical to
develop a communications program that would keep fans engaged
throughout the 8 weeks between registering for the contest and submitting their picks.
The solution was a weekly email campaign called “Countdown to the
Breeders’ Cup.” Contestants registered for the contest through an
online form that included seven segmenting questions. The final email
in the program, sent just hours after the rosters were finalized, contained a link to the actual “pick your horse” page.
To drum up interest in the contest, a combination of online, offline,
broadcast, and direct media was used. There were TV ads, emails with
“forward to a friend,” and signage displayed at racetracks.
The seven weekly emails to registrants created an overall clickthrough rate of 41 percent and an unsubscribe rate of 2 percent. Over
the 7 weeks, the campaign registered 23,407 people, of whom 87 percent, or 20,863 people, opted in to receive additional communications
and offers from the NTRA.
129
130
Stride Rite Shoes
Stride Rite Children’s Division had approximately 135 corporate retail
stores nationwide. Stride Rite’s marketing director, Julie Lehman,
turned to her account team at PreVision Marketing, LLC, in Lincoln,
Massachusetts, to create a strategy for filling the stores with customers
during the pre-Easter (traditionally a peak sales period) and April
(historically an off-peak selling season) time frames. The goal of the
strategy was not only to drive sales during these two periods, but also
to understand the impact of a second communication on the targeted
households reached.
The campaign consisted of two direct-mail pieces: a March peak
catalog and an April sandals postcard. The catalog was designed to display
the depth and breadth of Stride Rite’s product offerings and to drive
the key pre-Easter store sales. The follow-up April sandals postcard was
designed to increase store sales during the off-peak selling season by
focusing on a particular shoe product, sandals, to drive incremental sales
at the company’s retail stores.
For the campaign, PreVision used the Stride Rite customer database
for 70 percent of the households mailed and compiled prospect lists for
the remaining 30 percent to support newly opened Stride Rite stores.
The March peak catalog’s performance was measured over a 5-week
period during which a special offer featured in the catalog was valid
(“Buy two pairs, save 15 percent; buy three or more, save 30 percent”).
The April sandals postcard had no special offer, but its performance was
also measured for a 5-week period to ensure comparable results.
To test the effectiveness of each communication alone as well as the
incremental impact of the second communication, PreVision set up a
test design to enable the following comparisons:
• Compare those who received only the catalog to a no-contact
control group.
• Compare those who received only the sandals postcard to a nocontact control group.
• Compare those who received both mailings to those who received
only the catalog.
The no-contact control groups matched the audience selection criteria
of the mailed households, but did not get any of the communications.
To know who responded, each customer was tracked through name
and address capture at point of sale. These point-of-sale data were
loaded into the Stride Rite customer database and subsequently matched
to the outgoing mailing files. In this way, PreVision was able to identify which of the purchasers had received one communication, which
had received two communications, and which had received nothing.
The results, shown in Table 6-2, were very interesting. What do
these results tell us? First, both the catalog and the postcard as standalone pieces successfully increased both shopping rate and sales, resulting
in a positive return on investment (ROI). It is important to note that
the ROI for the March peak catalog was not as great as that for the April
sandals postcard, as the cost to produce the catalog was significantly
higher than the cost of the postcard.
The March peak catalog capitalized on a historically strong season for
selling shoes, as evidenced by a high natural shopping rate during that time
period. The catalog, which featured a large selection of shoes and showcased new styles, effectively increased the average basket size of those who
received the mailing compared to those who did not. In addition, the catalog successfully drove additional customers to shop at Stride Rite. The
April sandals postcard drove additional traffic during a typically slower
shoe-selling period but did not significantly increase average basket size.
The dual-contact strategy was also successful. The second contact
drove additional households to shop, especially if they had not shopped
Table 6-2 Stride Rite Results
Number of households in
mail group
Number of households in
control group
Lift in shopping rates
Lift in sales
Lift in average basket size
Return on investment
March peak
catalog vs.
no contact
April sandals
postcard vs.
no contact
Dual contact
vs. catalog
only
354,000
86,000
220,000
25,000
9.12%
15.16%
6.12%
1.28:1
20,000
20.93%
20.82%
NA
1.91:1
17,000
4.53%
3.27%
NA
1.22:1
131
132
after receiving the first communication. The lift in sales and traffic came
primarily from households that had not previously shopped. Because
the cost per piece for the postcard was low, there was a positive return
on the investment. The second communication paid off because it was
a cost-efficient piece.
Because of the success of this program, PreVision has recommended
dual-contact communication strategies during other selling seasons as
well, so that Stride Rite could maximize the effectiveness of its marketing dollars and communicate effectively with its customers.
Constructing an Email Test
If you are like most marketers, you have yet to do an organized email
promotion to your customer base. Here is how to go about it.
To begin with, you probably have a database of names and
addresses, but you probably have email names for only a few (or none)
of your customers. Your first job is to get the email names into your
customer records. As you know from this book, this process takes a
long time and involves revising all your forms and telemarketing
scripts to be sure you start capturing these email names, with permission to use them.
To jump-start the process, you can get opt-in email names appended.
You can get help with this. CSC Advanced Database Solutions, for
example, has access to a database of more than 50 million households
that have provided their email names to various vendors and given
written permission for their use for commercial promotions. You can
ask CSC or some similar vendor to run your house file against this list
and append your customers’ email names to your database records. This
will cost you about $250 per thousand, or about $0.25 each. (This may
sound expensive to you if you have not yet read Chapter 5, “The Value
of a Name.” After reading that chapter and doing your homework, you
will know whether spending $0.25 per name is worth it for you.)
There are dozens of other methods of gaining customer email
addresses. You should be sure that customer service and telesales always
ask for the emails and for permission to use them. There should be a space
for an email address on every form that customers fill out. Some companies have contests for their customers, with email addresses required for
entry.
Opt In or Opt Out
When you get customer email addresses, you should be sure that the
customers have given you permission to use them for communications.
There are two methods for ensuring permission:
• Opt In requires customers to check a box saying that they want to
hear from you. The result of using Opt In is often a 5 or 6 percent
customer email acquisition rate.
• Opt Out requires customers to check a box saying that they do not
want to hear from you. The result of using Opt Out is usually an
80 to 90 percent customer email name acquisition rate.
In both cases, customer privacy is protected. Customers can easily show
their unwillingness to hear from you. Don’t use Opt In. You will never
get enough customer email addresses to create a viable email marketing program.
Of course, you must make it easy for customers to discontinue further emails at any time by providing a “one click” option on every
single email you send to them:
• “To discontinue receiving any further promotional emails from
our company, Click Here.”
Should you get permission to sell the customer’s email name to others?
Of course you could. But do you want to? Many established businesses
find that the revenue gained from renting customer names is small compared to the negative perception of the company being in the name
rental business rather being a customer-friendly supplier. You will have
to decide what image you want to project to your customers.
Let’s assume that one way or another you have gotten email names
and permission to use them for 100,000 of your existing customers.
How do you go about developing a promotion test?
133
134
For direct mail, you would probably test two different packages,
sending them out simultaneously to 50,000 people each to see what the
response will be. You might even segment your file into 10 different
segments of 10,000 each, with slightly different copy, offers, and packages. After 6 weeks you would know which segment did the best job and
had the highest return on investment.
That is not at all the way email campaigns are run. In the first place,
you seldom, if ever, should send to your whole 100,000 at once. The
best plan is to divide your file into 20 different segments of 5000 each.
Then spend some time crafting the right appearance for your emails
on the customers’ screens. Figure 6-8 shows some samples of senders
and subjects.
Which of these grab you? How many of them would you delete
without reading them? I would say most of them. Only two caught
my eye: VAR Business and AOL. I subscribe to VAR Business and I
am a paying member of AOL, so I might open these two. In fact, I have
no interest in sending emails in five languages or in news about Computer Associates, so I would delete those two also. Figure 6-9 shows
another group.
There’s not much that looks interesting in this group either, except
possibly the message from Cyber9014 about processing online transactions. That might be a legitimate offer.
Figure 6-8
Email Sample 1
Figure 6-9
Email Sample 2
Lauren Arden sounds like someone’s name. It could be a legitimate
email that would be of interest. I decided to open that one. It was interesting. It began:
As a full Web-service provider, we are helping vendors sell their products directly to consumers. Therefore, prices for some of the products
are the lowest available. All 500,000+ discounted items can be found
by visiting www.ec-OurCity.com.
Let’s try another group (see Figure 6-10).
What stands out here is the Account Specialists with a subject Email
Centre. Why the English (rather than American) spelling of Centre? It
caught my eye. The email said:
The “@aol.com” part of your email address is owned by someone else
and is known as a Domain name. You will have datamktins@aol.com
as long as you subscribe and pay each month. You never really own your
email address, someone else does. You just rent it.
It went on to offer the firm’s services to help me own my own
email name.
135
136
Figure 6-10
Email Sample 3
The Valentine’s Day reminder also looked interesting. Notice that
dmckin324 is sending me two emails: one on credit cards and one on
background checks. He has destroyed his credibility by sending two
emails so close together.
What do these things tell us about emails?
• There is a lot of competition, particularly from really junk email.
• Junk email is much more widespread than junk direct mail.
Email is so cheap to send that any worthless smut purveyor can
afford to send millions per day.
• People look first at who sent the email and second at the subject.
They seldom look at the text of your emails at all, unless the email
address and subject pass a validity test.
• You need to give a lot of thought to what your email address should
look like.
• Then spend some time figuring out how to present your subject.
Settle on three different email addresses and three different
subjects.
• Once you are happy with these six, you can begin to worry about
what to say in the email. (I am reminded of experts at Ralston
Purina, who reportedly spent more than 6 months designing the
appearance of the perfect cereal box before they got around to
figuring what to put into the box!)
Designing the Text
Once people have opened your email and read your copy, what do you
want them to do? You have to make it easy, obvious, and compelling.
Here is how one merchant did it:
Click here to search our database with no obligation.
This takes you directly from the email to the merchant’s Web site,
where it offers its services (see Figure 6-11). This is very seamless and
compelling copy.
Figure 6-12 shows the text of the Valentine’s Day email. Each underlined section of text is an automatic link to a professional Web site. This
is the way to go.
Few email texts are this organized. Most of them make you work at
filling out forms and emailing them to someone, with a response generated in a few days.
Beginning Your Test
Your first step is to send 5000 emails in a head-to-head test of two different email addresses and one subject (2500 to each). You will know
within 48 hours which of the two produced the better result in terms
of click-throughs, responses, and sales.
Figure 6-11
Web Site for Designing Your Web Address
137
138
Figure 6-12
Text of Valentine’s Day Email
Now take the better of the two email addresses and try two different subjects with the next 5000 names. Save the best email address
and the best subject. Experiment not only with the email addresses
and the subjects, but with different texts and different days of the week
for your mailing.
Gradually, over the next couple of weeks, you can work your way
through your 100,000 customers, getting better and better response
rates as you go. You may start with a 0.5 percent response rate (12
responses out of 2500). At the end of 10 tests, you might have worked
your way up to 2 percent (50 responses out of 2500)—that is four times
better than the rate you started with. You might even do much better.
Many companies that use emails to market to their existing customers
have achieved response rates of 20 percent or more.
The key to success here is constant testing. You simply cannot do
this with direct mail. With email, if you do it right, you can achieve tremendous success in a very short space of time.
Summary
Communications to customers work. Used correctly, communications will
• Increase your retention rate and reduce attrition
• Increase cross sales and up sales
• Increase referrals
• Increase profits
The advent of email has given marketers a tremendously valuable
new communication channel that will revolutionize database marketing. The one-to-one communications that we talked about in the
1990s are being carried out in the new century. From now on, anyone who is not using emails will lose out in the competitive race to
build customer loyalty.
What Works
• Communication with customers through direct mail and
email
• Capturing email names with permission to use them
• Creating Web sites where customers and fans can register their
support of artists and products
• Online focus groups
• Real-time reports on the results of email communications
• Getting customers to fill out personal profiles online
• Personalizing emails and Web sites
• Testing emails on small groups to get the right email address
and subject
What Doesn’t Work
• Email blasts to unknown (and angry) people to get them to buy
legitimate products.
• Direct mail promoting low-margin products.
• Database marketing for packaged goods.
• Customer communications programs without the use of a control
group. You cannot prove that they have done any good.
• Trying to get your name eliminated from unsolicited email
commercials.
139
140
Quiz
1. What was the breakthrough solution in the Verizon case study?
a. Building a warehouse
b. Segmenting the data
c. Cooperation between IT and Marketing
d. Building a data mart
e. Sending mail first class
2. Figure the cost per response for an email campaign that costs
$0.05 per email and gets a response rate of 1.16 percent.
3. A bank tried two communications programs. One was to mutual
funds customers and was aimed at getting them to buy more
mutual funds. The second was to those who did not have a
home equity loan, asking them to apply for one. Which was
likely to get the larger response rate and why?
a. Home equity loan offer. It is easy to qualify for one.
b. Mutual fund offer. It is easier to get people to do more of
what they are doing than it is to get them to do something
else.
4. If RFM can double the response rate in direct-mail campaigns,
why would it not be equally valuable for email campaigns?
a. RFM does not work with email.
b. There is no control group.
c. Domain names are unavailable.
d. Emails are so inexpensive you can concentrate on
maximizing profits.
5. Propose a budget for your company for an online focus group.
List the two things that you would most like to get out of the
focus group.
6. Make a chart for a warranty expiration program, comparing the
costs of email warranty notification with telemarketing at a cost
of $5 per call. Assume that the emails cost $0.06 each. What
was the cost per order of each, assuming that the telemarketers
got 8 orders out of 100 and the email program got 3 orders out
of 100? What was the return on investment? Which method
would you choose?
7. List the four most important things to capture on a customer
profile completed through the Web.
8. What one factor, if it had been left out of the lighting
manufacturer’s test, would have ruined the program?
a. The customer rep
b. The lighting expert
c. The test group
d. The control group
9. In the case of the restaurant chain with the frequent diner card,
what did the program cost the chain over the 13 weeks?
10. Which is the most important item in an email promotion
program?
a. The offer
b. The timing
c. The copy
d. The subject line
e. The domain
141
C
H A P T E R
7
Customer Response
T
he best response to a promotion is a sale. Typically, you want customers to visit a store and buy your product or service. In the
absence of this fruitful response, until recently there were four other
ways in which customers could respond to a promotion. They could
•
•
•
•
Call on the phone
Send a letter through the mail
Send a fax
Send an email
These four methods are still used for both consumer promotions and
business-to-business marketing. They work well, and they will continue
to be an important part of direct marketing. But all four have a drawback: They are moderately expensive two-step procedures that require
additional effort to complete the sale.
In the last few years, Web response has been developed. It has many
advantages over these four methods. With Web response, the customer
goes to a Web site, gets more information about the product, and places
the order personally. This method swamps all the others in terms of
•
•
•
•
•
Reduction in fulfillment cost
Capture of data along with the sale
Immediacy of response information
Personalization and speed of response
Accuracy of the data captured
142
C u s tom e r R e s p on s e
Here is how the process works. In the promotional message, the customer is given the name of a Web site to contact. If the promotion is
an email, the Web site address can be contained in the message itself.
All the user has to do is to click where it says click here, and she jumps
to the site automatically. Once she arrives at the Web site, if the process
has been set up correctly, she sees a greeting: “Welcome, Susan” or
“Welcome back, Susan.” She also finds information about the product
that was featured in the promotion. At that point, she can
• Buy the product directly on the site or make a reservation (rather
than a sale)
• Get the names and addresses of local dealers who sell the product
• Talk directly to a local dealer
• Talk to a live operator
• Get detailed answers to her questions about the product or service
• Answer some survey questions designed to help her make a
decision about which product is best for her needs
• Complete a personal profile that provides her preferences and
some demographics
• Provide her address, email name, and credit card number
• Send messages to friends about the product or service
• Learn about other products sold by the supplier and order them
• Provide permission for further correspondence
In fact, in many ways, her visit to the site is better than a visit to a
local store. If the site is set up properly, it provides the supplier with a
wealth of information about the prospect or customer that would be difficult or impossible to get during a store visit or with any of the other
four response methods except the phone call.
“Welcome, Susan”
How do you manage to say, “Welcome, Susan”? This is one of the
secrets of modern database marketing using the Web. You begin with
a list of prospects who are to receive your direct-mail or email promotion. You assign a unique four-character code to each person on your
143
144
list. A typical code could look like this: R34Y. With a four-character
coding system, you can assign unique codes to 1.5 million people. A
five-character code gives you 52 million unique codes. You make your
Web site address as simple as possible: www.Nikkon.com/R34Y. When
Susan enters this address in her Internet browser or clicks on it, your
site functionality immediately checks the last four characters against a
lookup table that is stored in memory and is able to respond, “Welcome,
Susan.” Your software, with Susan’s permission, will leave a cookie on
her computer so that when she comes back the next time, you can say,
“Welcome back, Susan.”
If your promotion is via TV, radio, or print, the address will be different. You might tell the customer to contact www.Nikkon.com/BAR2.
The “B” might tell you that this is a mass communication promotion,
and the AR2 tells the Web site which of 42,000 particular ads on a TV
or radio station or in a periodical prompted Susan’s response. In this
case, of course, you cannot say “Welcome, Susan,” but you can know
exactly what prompted Susan to pay the Web site a visit.
Setting up a Microsite
Web response involves a Web microsite that is set up specifically to
receive customer responses to promotions. A microsite differs from a
regular company Web site in that it is constructed with a particular
group of promotions in mind. The setup cost is much lower. A major
company Web site, like www.marriott.com, costs millions of dollars to
create and maintain. (This site, by the way, is one of the best on the
Web.) Every company needs a main Web site, although most do not
need a site that is as complex and powerful as the Marriott site. What
we are describing in this chapter is a microsite that can be set up for a
one-time fee of $3000 and maintained for $400 per month or even less.
This microsite has a database associated with it on the same server. In
fact, the two are really part of the same system.
As an example of such an email connecting to a microsite, Figure 7-1
shows an email that I developed after one of my speeches. It was sent to
all the attendees who gave me their business card during the speech.
Figure 7-1
Sample Microsite
Any data entered into the microsite are stored immediately in the
database. Also, the database provides the lookup table for the PIN codes
that may be contained in the outgoing email or direct-mail promotions.
There are editing procedures stored in the microsite software that keep
erroneous data from being entered. As a result:
• U.S. zip codes are always numbers, and Canadian postal codes are
letters and numbers.
• Phone number area codes always match the address.
• Email names given by customers are always correct because the
responder gets a reward that is distributed by email.
• The site does not accept erroneous product codes.
• Dates, ages, and incomes are always within reasonable limits. For
example, customers’ ages are not listed as being over 120 years,
and birth dates are not later than today.
• Data are converted to the proper case, usually upper- and lowercase.
Finally, the microsite always has a seamless link to the company’s
main Web site.
145
146
Table 7-1 Comparative Media Results
Incoming costs
Data entry
Cost per 100,000
Time in weeks
Information gained
Possible errors
Interactivity
Personalization
Mail
Email
Fax
Phone
Web
$0.35
$0.60
$95,000
3
Medium
Moderate
None
None
$0.00
$0.60
$60,000
1
Medium
Moderate
None
None
$0.15
$0.60
$75,000
1
Low
High
None
None
$0.70
$3.50
$420,000
0
High
Low
Very high
High
$0.00
$0.00
$0
0
Highest
Very low
High
High
The beauty of Web response is that the customers do all the work.
The cost savings are tremendous. Table 7-1 gives the relative costs
per response.
With mail, email, or fax, it is necessary for a data entry clerk to read
and retype the information in order to get it into the company’s order
fulfillment and database system. There is always an error rate. Some companies report an error rate as high as 30 percent, and lots of data are not
captured—email name, phone number, or correct address, for example.
The time delay is significant. To keep costs down, data entry is done in
batches. As each batch is completed, it is sent electronically to the fulfillment and database system. In these systems, there is a limit on the amount
of information that you can ask for and reasonably expect to get.
The telephone is by far the best of the five systems. Until recently,
most call center operators typed their data into an internal software
system and sent the responses to the client’s fulfillment system in
batches. In modern database marketing, the operators have access to the
Web and can type the data directly into the client’s fulfillment and database system through the Web, so there is no delay. Products can be sent
out the same day. Operators can make suggestions and cross sales. They
can answer questions and coax a customer into buying more. You pay a
lot for phone service, but, when it is done well, you get a lot for it. For
most companies, it is still the best possible customer response system.
How the Web Can Trump the Phone
Where Web response comes into its own is in the sophistication it
makes possible and the very low cost. Once your Web microsite is set
up and running, the variable costs of each customer response are exactly
zero. We should note that we are not factoring in the setup costs here.
Web setup can be expensive, but it is much less expensive than call center
software. It is comparable to the elaborate arrangements that are necessary to receive, open, and enter data from phone, fax, or email.
The tremendous cost savings from use of the Web are possible only
if customers are willing to go to the Web to respond. Whether they do
so depends on how good the Web site is. If using the Web site is a
superb experience, word will get around, and customers will flock to it.
If the experience is annoying, unhelpful, uninformative, slow, or confusing, people will stay away in droves. For the present, do not expect
the majority of your customers to respond via the Web. The phone will
be the response mechanism of choice. But this is changing every day as
Web use grows. For the Web to be successful in database marketing,
your company is going to have to put a lot of effort into making the
experience satisfying for the responder. Here are some of the things
that you are going to have to provide:
• Personal greetings. Find a way to greet customers personally,
particularly business-to-business customers. As already discussed,
you absolutely must use cookies, unique site addresses, or PINs.
The goal is to make the Web site better than a live operator.
• Quick loading. Do away with flash pages that take a long time to
download. Most consumers today are still on modems, not
broadband DSL, and this will be true for some time to come. You
system has to be designed for your current customers, not for the
customers of the future. If you cannot make your current
customers happy, you may have no future.
• A site designed for the customer. Design your site with your
customers in mind. Put yourself in their shoes. Don’t say, “What
do I want to sell them on this site?” Instead say, “What would I
want to see if I came to this site as a customer who knows very
little about our company and its products and who is trying to find
information or get prices on a specific item?”
• Search boxes. Provide a search box prominently on every page.
Make that search box very informative. Nothing is more annoying
than going to a Web site looking for something and not being able
147
148
to find it. I went to the Hewlett-Packard Web site trying to find
toner for my HP Laserjet 4—the most common printer in the
United States. I couldn’t find it. Even if you don’t have what you
assume many people will be looking for, give them an answer. HP
could have said, “We don’t sell toner for our printers. Go to
www.Staples.com/HPToner.”
• Three-click rule. Use the three-click rule. Make it possible for
anyone to find anything on your site with just three clicks.
• Constant modification. Study the things that people are asking for
every day, and update the site so that it answers the questions
that are asked. That is what call centers do. If a number of
people ask the call center operator, “Is your product made in the
USA?” he will quickly ask around to be sure his answer is
correct. You have to design your Web site to do the same thing.
In designing and managing your Web response site, spend a few
hours a week sitting next to a call center operator and listening
to the questions and the operator’s answers. Your site has to be
that good.
• Pictures, lists, and diagrams. Capitalize on what you can do on the
Web site that a live operator cannot possibly do. You can have
pictures of things, maps, references, articles, lookup tables, and
diagrams. Incidentally, while I am at it, let me take a swipe at
Acrobat. Many Web sites use Acrobat software to display pictures
and text. It produces attractive results, but you cannot copy it or
download it easily. I prefer screen content that you can print
directly on your PC printer or copy directly into Word. Acrobat is
annoying and turns me off. Figure 7-2 shows the entire process in
a nutshell.
Of course, there are still millions of Americans who do not yet make
a habit of going to the Web in response to promotions. That number is
going down by thousands every day. Before you finish reading this chapter, several thousand additional U.S. households will have come online.
Today nearly everyone above the poverty level has a computer. Just
about everyone is using email at work or at home. While people are
looking at their email, it is an easy step to click on a Web site contained
in the email to respond to your promotion.
Figure 7-2
Customer Service Link through the Web
Customer
responds by
going to a Web
microsite
Welcome Susan.
Here are your
tickets to the…
Microsite is on
same server as
the database.
Fulfillment is
started
immediately.
Microsite
Database
Generating Web Response
I am often asked what the typical response rates to a promotion are.
There is, of course, no answer to this question. It depends on the audience, the offer, the package, the copy, the timing, and a host of other
factors. Some promotions to customers have generated response rates
of 40 percent or more. Most promotions to prospects are lucky to
achieve a 1 percent response rate. An important aspect of promotion
response is the method selected for receiving the response. Is it mail,
phone, email or Web? As this book demonstrates, Web response is the
most cost-effective and powerful method. Let’s take a look at a couple
of cases to demonstrate this point.
Direct-Mail Response
An automobile manufacturer designed a credit card for its automobile
customers. It created two different cards: a Titanium card with a $100
annual fee and a Gold card with no annual fee. The Gold card included
a $50 reward package and an initial APR of 3.9 percent.
The direct-mail promotions were sent only to the owners of the
company’s cars, not to the general public. The automobiles were
149
150
designed for and sold to an upper-income segment that appreciated their
performance and style. The owners were loyal to the brand. The promotions took place at a time when the credit card market had been saturated with offerings for the previous four years. Few cards were
generating significant response rates.
Overall, there was a surprising difference in the response to the two
cards, as shown in Table 7-2. From these numbers, it is clear that the
Titanium card was a loser. The acquisition cost of $297 per account was
really uneconomical. You can’t subtract the $100 annual fee from this
number because the fee will be paid only at renewal time, 1 year hence.
At that time, if this card is like most cards, less than 50 percent will pay
the $100 to renew.
The Gold card, on the other hand, produced an unusually good
response rate. Since the overall industry average acquisition cost for
credit cards is well over $100, this automobile company made excellent
use of its customer database to promote these cards, as its cost per
account was only $57.
The response to the cards was by phone or mail. No opportunity for
Web response was offered. This could have been a serious mistake.
Direct Mail versus Email
A major manufacturer of business-to-business computer products created an innovative Web site that was designed to
• Teach IT managers about the value of the company’s services
• Get them to register
• Profile them for the sales force
Table 7-2 Card Response
Low response
High response
Total response
Cost/response
Approval rate
Cost/account
Titanium
Gold
0.15%
0.40%
0.25%
$317.50
107.00%
$297.25
1.01%
4.71%
1.60%
$49.96
87.22%
$57.28
A creative direct-mail piece was sent to 186,000 IT professionals
drawn from lists of customers and rented lists of computer magazine
subscribers. It directed people to go to the Web site. Parallel to this, an
email campaign was launched to 56,000 IT professionals. The email lists
were primarily lists of trade show attendees. The results comparing the
various lists used, shown in Table 7-3, were quite interesting.
What Table 7-3 shows is that the email promotion did very well
against the direct-mail promotion. The response rates were lower and
the percent registering was much lower, but the cost per registrant was
almost half that of direct mail. What can we conclude from this?
• Direct mail is not dead. It will continue to be a useful and
effective direct marketing technique.
• Email gets lower response rates, but it produces results at far
lower cost.
There is one really powerful lesson from this case that does not show
up in the numbers presented so far. That lesson concerns the responses
that came from an unknown source, as shown in Table 7-4. This is
where the company was a real winner.
Table 7-3 Cost per Registrant
Cost per piece
Number sent
Low response
High response
Total response
Percent registering
Cost per registrant
Direct mail
Email
$0.76
186,361
0.26%
4.03%
1.42%
1.01%
$75.25
$ 0.14
53,250
0.00%
7.79%
1.04%
0.29%
$48.28
Table 7-4 Responses from an Unknown Source
Number of log-ons
Number registered
Cost per registrant
Gross revenue
Marketing costs
Net revenue
Direct
mail
2,639
1,885
$75.25
$942,500
$141,846
$800,654
Email
553
153
$48.28
$76,500
$ 7,387
$69,113
Unknown
source
1,288
882
$00.00
$441,000
$0
$441,000
151
152
Where did all of these “unknown source” registrants come from?
Analysis of this situation tells us a lot about modern direct response
techniques.
The Value of Web White Mail
In a traditional direct-mail campaign like the credit card example, the
responses arrive via a toll-free number. When you call this number, an
operator asks you for your source code. Good operators can usually
worm this number out of the respondents, even though they may not
have it with them when they call. When the operators cannot determine the source, they dub the source “white mail”—responses from an
unknown source.
On the Web site, it is different. In this example, responders were supposed to enter their PIN when they came to the site as a result of an
email or a direct-mail piece. If they did not remember their PIN, they
were invited to visit the site and register anyway. As you can see, 1288
of them visited the site, and 882 registered.
We will never know whether these 1288 people came as a result of
the email campaign, whether they came as a result of the direct-mail
campaign, whether they stumbled on the site through the recommendations of a friend or associate at work, or whether they came because
of a combination of these reasons. But that points up a difference
between live phone operator response and Web response: the fact that
an entertaining Web site with a relevant quiz can lead people to visit the
site to see what it is. There is no commitment. You can be anonymous,
and disappear if you are not interested.
On the other hand, practically nobody calls a toll-free number just
to see what it is like. There it is tough to be anonymous. You know
you will be talking to an operator. You don’t want to waste her time.
Therefore,
• Web response will generate a lot more white mail than phone
response.
• This response can be the most valuable output of the entire process.
Selling Trucks to Existing Customers
Isuzu makes midsized commercial trucks that sell for about $33,000
each. Recently, the truck marketplace was moving very slowly. Furthermore, at that time, Isuzu had no new truck models to offer with which
it could beat the competition. As a result, Isuzu had to rely on better
marketing and improved ability to target prospects and customers.
At the time, Isuzu’s marketing budget had been cut back. On the
bright side, however, American Isuzu Commercial Vehicles had just
completed two major database projects using an updated customer database created by msdbm in Los Angeles. The first project was to profile
customers, and the second was to model customer purchase behavior.
Based on these models, a numerical scoring methodology was developed to indicate purchase behavior. Scores were then grouped into
three ranges indicating a high, medium, or low likelihood of purchase.
If the methodology was accurate, then customers who received a high
score would be more likely to purchase in the near term than customers
who received a low score. It was also felt that the company was more
likely to be successful at selling a new vehicle to a current customer than
at trying to convert a prospect.
Msdbm developed a scoring methodology to predict the behavior of
Isuzu Commercial Truck owners. The model predicted the propensity
to purchase based on four variables:
1.
2.
3.
4.
Number of people employed by the business
Vocation—e.g., landscaping, electrical contractor, towing, etc.
Time since last purchase
Total number of trucks owned
Using the model scores, the customers were divided into three
groups based on the likelihood of their purchasing a new truck:
1. High probability
2. Medium probability
3. Low probability
153
154
Isuzu decided to use the models to create a mailing designed to drive
selected customers to Isuzu dealers. The goal of the mailing was to
• Test the model to determine whether the scoring methodology
could correctly separate the customers into the three segments.
• Use the same creative material for all segments—low, medium,
and high scores—in order to get a valid test.
• Provide a mechanism for collecting and updating data and offer a
purchase incentive.
• Sell 150 trucks.
The data in the legacy system had many flaws. They were over a year
old. Much of the contact management history had not been recorded.
The majority of the customers were small vocational companies, such
as Joe & Jim’s Landscape Company in Keokuk, Iowa. The vocation of
many other companies was difficult to identify. The data used for the
model were put together from multiple sources of data: R. L. Polk,
InfoUSA, and in-house transaction files.
There was an additional problem of determining who the real driver
of the truck was. In a typical situation, the sale of a truck might have
been financed by GE Capital, then the truck was delivered to Ryder
Truck, leased with a new truck body to Joe & Jim’s Landscape, then
driven by Mac McKenzie. Who is the decision maker for the purchase
of a new truck?
What Was Sent to the Customers
Msdbm created a direct-mail package that included a cover letter, a faxback form asking specific purchasing questions, a four-color product
information sheet highlighting the features and benefits of the most
popular truck model (N Series), and a purchase certificate for an additional gift if a truck were purchased. The completion of the fax-back
purchase information form guaranteed all respondents a gift of a briefcase, while the purchase of a new or used vehicle guaranteed the buyer
a set of seat covers and floor mats for use in the vehicle. Respondents were
given only 60 days to take advantage of the offer. The total quantity
Table 7-5 Trucks Sold
High score
Medium score
Low score
Control group
Total
Group
totals
Companies
buying
Buying
rate
Trucks
sold
1899
1855
1890
520
6164
97
32
15
3
144
5.10%
1.70%
0.80%
0.60%
2.30%
107
34
16
3
160
mailed was 6000. In addition, 520 companies were set aside as a control
group to monitor the success of the promotion. These 520 companies
did not get the promotions.
This test mailing proved that the scoring methodology that had been
developed was highly accurate; the results are shown in Table 7-5. As a
result, Isuzu used the same methodology to score the entire customer
file for a follow-up mailing. This gave Isuzu the ability to target key
customers and ensure the sale of a vehicle at very little cost, compared
to normal prospect acquisition costs.
The marketing and sales departments were initially dubious about
the mailing. Based on their previous experience, they had expected to
sell only 85 trucks to 6000 prospects. After they saw the results, they
could see that their future success was going to be tied to database marketing. Being able to accurately predict which companies were likely to
buy, so that they could direct their marketing and sales efforts at those
companies, proved to be a powerful advantage.
As a result of the test campaign, the company sold 160 trucks, for a
total of more than $5,600,000, with an investment, including the model,
printing, mailing, and generation of creative material, of $80,000 (see
Table 7-6). The industry estimates that, in general, it costs a manufacturer about $1000 in marketing and sales effort to sell one midsized truck.
Using the database, Isuzu was able to sell the trucks for half that amount.
Web-Based Employee Recognition Program
ScotiaApplause was designed by the Carlson Marketing Group to be a
platform that would support a customer-centric employee recognition
155
156
Table 7-6 Overall Truck Promotion Results
Description
Results/response
Total number of trucks sold in 60 days
Overall customer sales conversion
Overall sales conversion on total trucks purchased
Overall forecasted sales conversion
High-score sales conversion on trucks purchased
Medium-score sales conversion on trucks purchased
Low-score sales conversion on trucks purchased
Cost of program
Program cost per vehicle sold
Control group (520)—3 sales
Total delivered mail quantity
160
2.1%
2.8%
1.5%
5.1%
1.7%
0.8%
$80,000
$500
0.5%
5,644
program for 20,000 branch employees of Scotiabank, Canada’s third
largest bank. The purpose of the Applause program was to promote,
recognize, and reward certain customer-centric behaviors defined by the
bank. To focus and articulate the brand vision, a statement of the bank’s
core valves was developed. The core purpose was considered critical in
sustaining the bank’s position as a leading international Canadian bank.
The core purpose statement was
To be the best at helping customers become financially better off by
finding relevant solutions to their unique needs.
Carlson and Scotia management designed ScotiaApplause to encourage employee behaviors that demonstrated the desired core values. The
Applause Principles for employees were identified as
•
•
•
•
Respectful
Straightforward
Insightful
Spirited
Background of the Program
Historically, employee incentive programs at Scotiabank had been
used sporadically and had had a strong emphasis on driving volumes.
Many of these programs encountered difficulties ranging from lack of
implementation to inequitable competitive groupings. Most of them
lacked a strong reinforcement for building customer relationships. The
worst characteristics of these programs were the following:
• They were strictly sales incentives.
• The areas where the bank’s budget was biggest, rather than
customers’ needs, got most of the attention.
• They offered few benefits for the sales and support staff.
• They did not benefit most employees, and as a result they did not
lead to discernible changes in employee behavior.
• They lacked consistency in the communications delivering
recognition or rewards.
• They did not address the primary employee behavioral problems.
From a review of past programs, Carlson concluded that a program
designed around recognizing and rewarding leading indicators of performance activities and behaviors, as opposed to lagging indicators or
results, was required.
Strategy and Design
Many Scotiabank branch employees had been with the bank for 15 to
25 years. Securing their buy-in for a new incentive initiative was not
easy. The core idea of the program was that establishing a recognition
system that rewarded employees for adhering to the identified core values
would lead to an increase in customer service, resulting in customer
retention and growth while minimizing attrition.
ScotiaApplause was designed to be completely Web-based, with
employees accessing the Web site from work or from home. By means
of emails and intranet connectivity, employees could electronically submit
both online recognition and personalized certificates. Certificate nominees and recipients were eligible for monthly reward draws for Applause
points, which were redeemable for reward merchandise.
ScotiaApplause gave the bank
• An ability to offer district vice presidents and branch managers a
motivational program
157
158
• A recognition program that was integrated among the various
channels—call centers, the branch network, and management
• Flexibility by recognizing both team goals and individual
achievement
• Success with online training
• Real-time reporting for management and staff
Results after 6 Months
The initial goal was to get employees to register for the program. The
target was to get 75 percent to register by the end of the fiscal year. Six
months ahead of schedule, the program had achieved 91 percent registration, with about 70 percent of those who had registered being active
in the program. The goal for certificate issuance was 660 per week. In
fact, the program actually achieved 2844 per week.
• More than 100,000 recognition certificates were issued.
• More than 7000 rewards were ordered.
• Desired activities and behaviors were rewarded, and the result was
an increase in these activities and behaviors.
• There was consistency in communication channels and messages.
The program received significant recognition throughout the bank.
It rapidly became the bank standard for employee recognition and
rewards. The success of the program contributed to the bank’s being
rated the number one bank in Canada in overall quality of customer
service by the Customer Service Index (CSI) survey, published by Market
Facts, a leading U.S.-based market research firm.
Moving a Loyalty Program to the Web
A major long-distance provider asked KnowledgeBase Marketing, an
integrated marketing solutions company, to develop a new loyalty database and redesign a rewards program for the company’s highest-value
customers. The goal of the program was to prevent churn and provide
cross-selling opportunities. For 9 years, the long-distance provider had
maintained a customer retention program on a mainframe, operating
in batch mode. In Phase I, KnowledgeBase Marketing upgraded the
system to a true real-time transactional database. Customers now could
redeem loyalty points in real time through a call center, through an
automated telephone system, or through the travel division. In addition, the new system provided the ability to offer more exciting rewards
efforts, cross-selling opportunities, promotional tracking, and
enhanced customer interaction with the company.
How the Loyalty Program Worked
Reward program members accumulated points through monthly longdistance phone usage. In addition, they were awarded quarterly and
anniversary bonus points, as well as points for responses to special
promotional campaigns. The accumulated points could be redeemed for
invoice credits, travel options, and gift certificates for additional products and services.
Objectives of the Shift to the Web
KnowledgeBase’s real-time database was a very effective system for this
long-distance provider. However, there were opportunities to improve
costs and deepen the customer relationship by bringing the loyalty initiative directly to the rewards membership. Therefore, in Phase II,
KnowledgeBase Marketing created an online redemption functionality
for the loyalty platform, with the following objectives:
• Provide a Web-enabled interactive interface for the rewards
program through the company’s own Web site, crossing through
firewalls, to the KnowledgeBase loyalty database.
• Deliver the new system in 30 days.
• Increase retention and cross-selling opportunities with the new
system.
• Reduce the operating costs of the loyalty program.
159
160
With the new system, customers could go online to obtain information, review their transaction history, and redeem loyalty points.
Customers received email confirmation of their redemption seconds
after the transaction, further creating a “live link” with the customer.
Analysis of the rewards program, such as determining the type, frequency, and medium (Web, IVR, call center, etc.) of redemptions, is
performed using a reporting database into which customer redemptions and other data are fed. IVR, of course, is Interactive Voice
Response, the process of pushing buttons on a telephone keypad to get
the information or the person you are seeking from an automatic
phone answering system.
In the first week after the program was launched, the company had
a 40 percent increase in point redemption. In subsequent weeks, Web
redemptions grew by 5 to 10 percent. Web redemptions, of course,
replaced phone redemptions. The average cost of a phone redemption was about $6. The marginal cost of a Web redemption was only
a few cents. In the first year of the Web redemption program, the
long-distance provider’s costs for maintaining the loyalty program
were cut in half. Churn was reduced, and cross-selling initiatives were
enabled through customer-friendly Web marketing efforts assisted
by KnowledgeBase Marketing.
Summary
Web response beats phone, mail, email, and fax in
•
•
•
•
•
Reduction in cost of fulfillment
Data captured along with the sale
Immediacy of response information
Personalization and speed of response
Accuracy of the data captured
Web response involves the creation of a special microsite where customers can be recognized and greeted by name, using a PIN or a cookie.
Direct-mail promotions usually get better response rates than email
promotions, but the cost per response may be higher.
The Isuzu case showed that models can correctly predict which
business customers are most likely to respond and which probably won’t.
The Scotiabank Applause program shows that internal Web programs can be highly effective in reaching employees and obtaining
employee participation, even with a mature group.
The KnowledgeBase program showed that long-distance customers
will respond productively to an online program, saving the cost of
phone response.
What Works
•
•
•
•
Getting your Web site to do things that a live operator cannot do
Using the three-click rule and having a search box on every page
Reaching employees through an internal Web program
Online response
What Doesn’t Work
• Flash pages that take more than 5 seconds to download
• Searches that don’t provide what customers are looking for
• Sites that are designed to sell, rather than being designed to
provide what customers want
• Mainframe legacy databases in support of online operations
Quiz
1. Why do Web response campaigns produce more white mail than
phone response campaigns?
a. People are curious about the offering.
b. Operators cannot worm source codes out of callers.
161
162
2.
3.
4.
5.
6.
7.
c. On the Web you can’t easily capture source codes.
d. All of the above.
Which of the following is not suggested as a way of improving a
Web site?
a. A high-tech flash page
b. Personal greetings
c. Search boxes
d. The three-click rule
What is the cost per registrant of an email campaign that costs
$0.11 per email and produces a 0.67 percent click-through rate
and a 48 percent registration rate?
What was the essential lesson in the Isuzu truck case?
a. You can sell a lot of trucks through direct mail.
b. You can accurately predict response rates.
c. You need a customer database.
d. Truck buyers are business-to-business customers.
e. The control group beat the tests.
The Scotiabank Applause program proved all but which of the
following?
a. The power of the Web.
b. That long-term employees will buy into a new program.
c. That employees will work to receive rewards.
d. That employee rewards program work.
e. None of the above; it proved all of them.
In the KnowledgeBase case study, the key goal was
a. To create a profitable Web site
b. To increase customer transactions
c. To deliver a new system in 60 days
d. To reduce call center costs
e. To reduce churn
Which of the following was not suggested as something to have
on your Web site?
a. Pictures
b. Maps
c. Articles
d. Diagrams
e. Video clips
8. Which has the highest rate of errors as a response mechanism?
a. Email
b. Fax
c. Phone
d. Web
e. Mail
9. Which has the lowest cost per 100,000 as a response mechanism?
a. Mail
b. Phone
c. Email
d. Fax
e. All have the same cost
10. Which has the highest level of interactivity as a response
mechanism?
a. Phone
b. Web
c. Email
d. Fax
e. Mail
163
C
H A P T E R
8
Marketing to
Customer Segments
To develop a relationship program, you still have to put individuals into
groups, and develop products and strategies that will keep them loyal.
That, in my opinion, is where companies have the most difficulty. Even
when you say to them, “I can help you to identify your key customer segments.” They respond, “Well, great. But tell me, what to do with them
once they are identified. How do I manage each segment?” Marketers are
not yet sophisticated enough to know what to do with the information.
—Stephen Shaw, Canadian Direct Marketing News
S
ince one-to-one marketing is seldom possible, most successful direct
marketers aim their marketing efforts at segments. A segment is a
group of customers or prospects that you have identified for marketing
purposes. In your mind, you think, “What are these people like? What
would appeal to them and get them to buy? How can we track their
response to our promotions?” Then you set about getting the information that you need in order to define the segment and design marketing
programs just for it.
Creating marketing segments is an art, not a science. Expert marketers spend a lot of time dreaming up segments and consulting their
database to see if these segments really exist. Then they spend time
coming up with marketing programs and testing them on the segments.
To determine whether they are successful, they always carve out control
164
M a r k e t i n g to C u s tom e r S e g m e n ts
Segment Marketing versus One-to-One Marketing
The idea in one-to-one marketing is to store in a data warehouse
enough information about each customer to enable you to
develop an individual marketing program designed just for that
customer. True one-to-one marketing is real time: If a customer
buys something this morning, your communications to that
customer this afternoon will reflect that fact, along with
everything the customer has purchased since the beginning of the
relationship with you, plus what you know about the customer’s
demographics. You don’t have five marketing programs; you have
2 million marketing programs for your 2 million customers. It is a
very nice idea. So why isn’t everyone doing it? Because, in most
cases, the lift in sales that you get from one-to-one marketing
does not pay for the cost of the warehouse and the
communications. There are very, very few situations in which real
one-to-one marketing is economically profitable.
Segment marketing, on the other hand, is relatively easy to
achieve. You still need a database of customers that includes
their demographics, preferences, and transactions. Your next
step, however, is to divide customers into segments on the basis
of profitability: Best Customers, New Customers, Lapsed
Customers, and Occasional Customers. You develop a marketing
program for each segment. The actual communication with each
customer can be personalized using the data on that customer
that you have in your database, but the segment determines the
overall approach. The cost of segment marketing is a fraction of
the cost of one-to-one marketing.
groups, members of the segment who do not receive the promotions
that are sent to all the other members of the segment. Then they measure their success by comparing the purchases made by the members of
the segment to whom the promotion was sent against the purchases
made by the control group. To determine the return on investment,
165
166
they have to take the incremental revenue gained from the segment to
which the promotion was sent and subtract it from it the cost of the
promotion and the rewards given to the members of that segment. Many
times the return on investment is negative. In this chapter, we will begin
with the process of creating a segment: coming up with the idea and finding the data to measure whether the segment is performing as you
intend. Then we will look at some case studies involving marketing to
segments. So, let’s begin with getting data about segment members.
Two Kinds of Data
There are two kinds of data: demographic and behavioral. Demographic data include age, income, presence of children, lifestyle, geography, type of housing, and other such things. In the case of
business-to-business files, the demographic data are SIC code, number
of employees, annual sales, and so on. Behavioral data usually consist of
transactions—responses, purchases, redemption of certificates—plus
preferences and survey responses. In business-to-business marketing,
transaction data are usually very easy to acquire, since most businesses
have accounts with their suppliers. With consumers, the data are often
difficult to get your hands on, particularly if the product is purchased
at retail stores or through dealers.
There are two ways to obtain demographic data: You can have them
appended to your database records by an outside source such as
Experian, or you can get the information directly from the customers
by asking them to fill out a survey or a profile. The profile method has
become so important that a whole chapter, Chapter 14, has been
devoted to the subject.
To market to segments, you need a database that includes transaction history. The difficulty is in capturing the data. Many retail stores
try to do this with a proprietary credit card. All large department stores
issue cards, which they hope their customers will use. In fact, only the
most determined customers use these cards. It is a nuisance to carry
around 10 or 12 store cards in addition to your MasterCard, Visa, or
Figure 8-1 iDine
Amex. The solution to this problem is to get customers to register their
bank credit cards with you, so that when the customers use these cards,
you will know that they have made a purchase.
iDine (see Figure 8-1) has made a successful business out of this idea.
Members register their credit cards with iDine. Then, whenever they
go to any of the 7500 restaurants in the country that are associated with
iDine and pay for their purchases using one of these cards, iDine finds
out about it and either gives them a credit on their credit card or gives
them airline miles. The customers don’t have to say to the restaurant, “Be
sure to let iDine know that I am eating here!” The process is so simple
that it is amazing that more retailers have not caught onto the idea.
Reverse Phone Matching
In the absence of a registered credit card to identify their customers—
and because it is often more effective—many retailers use reverse
telephone matching to capture customer data. The Sports Authority,
the largest full-line sporting goods equipment retailer in the United
States, is a good example.
167
168
Retail sales are often driven by the calendar. People are in a buying
mood before Christmas. The mood is entirely different a month later.
The Sports Authority, which operates 198 full-line sporting goods superstores in 32 states across the United States, wanted to respond to existing
and new customers in key markets during the busy holiday shopping
season. Like most retailers, The Sports Authority gathered information
about customers at the point of sale. However, by the time information on a customer made its way to the company’s marketing department
for use in follow-up or promotions, at least a month had elapsed. To
capture repeat business before the 4-week holiday season ended, Jeff
Handler, senior vice president, advertising and marketing, challenged the
company’s customer relationship management (CRM) software vendor,
Harte-Hanks, to shorten this time to 1 week.
The goal was to identify the customers, and then, using print-ondemand functionality, react to those customers with an appropriate
direct-mail offer, all within 1 week after any store visit. Few retailers in
America have ever accomplished such a feat. The Harte-Hanks customer
contact strategy, dubbed “Retail Daily Sales Builder,” involved analysis
of point-of-sale data, interaction with the full customer database, and
creation of customized marketing communications.
The Sports Authority centers its holiday promotions around the first
week in December. For this program, The Sports Authority focused on
its in-store customers, as opposed to Web or catalog customers.
The Objective: Repeat Business
To make the most of the holiday season, The Sports Authority uses
both mass marketing and targeted, focused direct mail to bring shoppers into the stores in December. Retail Daily Sales Builder was
designed to cost-effectively bring customers back one or more times
before Christmas, a time period when repeat business may be difficult
to achieve because customers have many choices. Without special incentives or relevant messages, they may not return to stores where they
have already spent money for some of their holiday purchases. The key
is to provide compelling offers in a timely manner. Without a truly
sophisticated system, compelling and timely offers are not possible. The
Sports Authority’s goals were to get a printed, customized direct-mail
piece into customers’ and prospects’ hands within 1 week following a
store visit, and to reward early purchasers during the holiday season
The most difficult elements of the project were getting the program
in place by November 30, given that the starting point was early
November, and then being able to quickly turn around a second offer
to customers and prospects. Part of the process involved matching the
in-store customers to their mailing addresses. The team used a reverse
telephone number appending process. Each clerk at each point-of-sale
station captured the buyer’s telephone number as part of the transaction.
Every night, these data were transmitted from each of The Sports
Authority’s stores to the central database, where the phone numbers
were matched to mailing addresses using reverse appending. With its
Retail Daily Sales Builder process, The Sports Authority and the HarteHanks team were able to do that within 1 day of the purchase. Many
of the names and addresses were already in The Sports Authority’s
customer database. Other names and addresses came from rented telephone appending lists.
The Offer
The day after any purchase, Harte-Hanks was able to identify the name
and address of almost every person who had provided a phone number the
day before. It was then able to print a targeted direct-mail certificate
and mail it so that it would arrive in the shopper’s mailbox within 1 week
of the shopper’s visit to the store. The project started in November with
the mailing of a holiday gift guide to all existing retail customers near
each store. In certain key markets, there was some targeted prospecting
as well. People who made a purchase based on the gift guide would
receive a second standard postcard offer within 1 week.
The Sports Authority says that approximately 5 percent of the shoppers who received the initial gift guide came to the stores and used the
certificates included in the guide. The follow-on mailing to anyone who
responded to the first holiday mailer had double-digit response rates,
essentially tripling the industry norm for response rates and, most important, bringing those customers back to shop again before Christmas.
169
170
Why did this program work? Because of its recency and a solid offer.
The customer who is most likely to respond to any promotion is the
customer who has made a purchase most recently. The increased
sales were so significant in comparison to the cost of the program
that The Sports Authority made it a permanent part of its marketing program.
Creating the Segments
Once you have captured the customer transaction data, you need to
divide your customers into some sort of usable segments so that you can
create a marketing program for each. How do you set up the segments?
This is a problem that separates the good marketers from the indifferent
ones. There are dozens of ways to do this. For example:
• Senior citizens, college students, families with children, empty
nesters
• High spenders, occasional shoppers, lapsed customers
• Profitability segments
• RFM (recency, frequency, monetary analysis) segments
• SIC groups (for business-to-business marketing)
Bringing Them Back
The School of Continuing Studies at the University of Toronto is a selfsustaining academic division of the university that welcomes about
15,000 adult learners each year and offers courses in three program
areas: business and professional studies, arts and humanities, and English
as a second language.
Recently, the school had to replace its existing registration system.
This system had been in place for more than 12 years, and, although
it could generate business reports and track registrations, the actual
customer data were locked within its tables. The new system was
Oracle-based. The school’s director saw this as an opportunity for the
school to take advantage of its customer data and employed Terri
O’Connor as the school’s first database marketer.
Making Sense of the Data
Terri put the customer records from both the old and the new registration systems into a separate database. In order to make sense of the
data, the records from the two systems were merged to create a unique
customer record that provided a profile of the customer. During this
process, Terri used RFM to help summarize the customer data.
The frequency indicator provided the school with its first database
surprise. The administrators had always known that they had a significant number of long-term customers, but they were very surprised to
learn that some customers had purchased more than 50 courses in the
previous 12 years. Terri termed these the high-value (HV) customers
and started to track their retention rates and the types of courses they
were purchasing. The breakdown is shown in Figure 8-2.
Recency and LTV
The school used several methods to distribute its course calendars. One
of the most effective methods was to mail them directly to customers
Figure 8-2 Currently Registered HV Customers by Start Date and
Program Area
70
60
50
40
30
20
10
0
88 989 990 991 992 993 994 995 996 997 998 999 000 001 002
1 1 1
1
1 1
1
1 1
1 1 2 2
2
19
Writing
Strategic leadership
Science
MNT
Liberal studies
Languages
French
ESL intensive
Dispute resolution
BPE DLP
BPE
171
172
who had registered within the previous 2 years. When Terri started to
assemble the first house list using the new database, she assigned a
source code to each customer for tracking purposes. The addition of
the source code meant that the list and the individual customer cells could
be tracked separately. She reviewed the customer records and was able
to locate and include a group of customers who had a high lifetime value
(LTV) but had not purchased courses for 3 or 4 years. Normally this
group of customers would be excluded, since it is too costly to mail calendars to customers who haven’t purchased in 3 or 4 years. But by targeting only the customers in this group who had a high LTV, Terri was
able to keep the mailing costs manageable while she attempted to reactivate their business. The result was that 8 percent of those who were
targeted purchased courses, and their purchasing habits were as high
the second time around as they had been originally. The reactivated HV
group bought an average of three courses each per year compared to
the school’s overall average of 1.45.
The impact of recency on course purchases is obvious in Figure 8-3.
By assigning RFM indicators to the customer base, Terri was able to
find new revenue from “lost” customers and to gain some valuable
insights into the school’s best customers.
Marketing to customer segments by direct-mail promotions is relatively easy. Going beyond that by appointing segment managers and
Figure 8-3 Sales Rates for the House List Calendar Mailing
50%
Reactivated HV Grp
40%
30%
20%
10%
0%
1mo.
<12mo.
12mo.
<24mo.
Months since last registration
<48mo. Cal Req Alumni Test
NR
NR
never registered
managing segments on a daily basis is much more complicated and
seldom profitable. Let’s take a look at one enterprise that succeeded.
Targeting Customer Segments
While many companies have failed at managing customer segments on a
day-to-day basis, National Australia Bank (NAB) succeeded at this task.
The lead franchise in NAB was National, a Melbourne-based bank
serving more than 4.5 million customers. With the help of Beth Ann
Konves of the Teradata Division of NCR, Fernando Ricardo, head of customer knowledge management at the bank, identified six major customer
segments composed of individuals and businesses. Each segment was managed as a global business. The system, called National Leads, identified leads
within the bank’s database based on significant changes (events) in a customer’s transaction pattern, such as balance changes, large transactions,
or loan renewals. The system also calculated each customer’s propensity
to buy a product or to respond to an offer. The bank’s database was maintained on an NCR Teradata computer using Teradata’s CRM analytical
software, which was used for the National Leads system.
The National Leads system prioritized events and alerted the appropriate bankers each morning so that they could take appropriate action.
The system also served as a communications gatekeeper, managing the
frequency, content, and channel used for customer interactions during
marketing campaigns. The goal was to ensure that customers were contacted with meaningful opportunities at the right times through the
right channels.
The six major customer segments used by the bank were common to
many other banks and financial institutions:
•
•
•
•
•
•
Custom business
Packaged business
Agribusiness
Private
Premium
Retail
173
174
Each segment was managed by a vice president. Every National customer was assigned to one of the segments on the basis of a financial
needs analysis and profitability assessment. Within each of these major
segments, further refinements were made based on current profitability,
the bank’s share of the customer’s wallet potential, and the customer’s
demonstrated financial needs. As a result of this analysis, there were
more than 200 subsegments that were used for target marketing. The
system worked this way:
• The Teradata warehouse was updated with an average of 2.2
million transactions per day.
• Every night, queries “trawled” the warehouse to search for any
unusual changes. The queries involved complex logic, scanning
hundreds of millions of rows in the database and joining from six
to ten tables at a time.
• Once a month, each customer was scored, using a model that
predicted the customer’s propensity to purchase various products
and propensity to respond to product offers. The best leads were
selected and sent to the bankers.
• Every night 250 communication vectors combining the events
detected and the propensity predictions were run. The software
recommended the action to be taken via ATM, email, mail, or
leads to call centers, branches, or package business bankers, who
followed up with a phone call or personal mail.
• The system captured all feedback and responses, measuring them
against the opportunities developed. Each offer was placed in a
location in a three-dimensional “cube,” with the three axes being
the segment, the channel, and the offer. The system moved
customers from one location in the cube to another based on their
response to offers.
In a typical scenario, the system might direct a banker to contact a
customer with an offer, such as a financial planning appointment or the
renewal of a loan. Armed with this lead, the banker would contact the
customer to gather financial profile information, using the standard
financial needs assessment process developed by the bank. Bankers were
given incentives to gather pieces of profile information every time they
spoke to their customers. The bankers typed the collected data into a
customer needs record in the database, using a terminal on their desk.
The way it worked was this: A customer might tell a banker that he
was currently renting, but that he planned to purchase a home in about
9 to 12 months. The banker would enter this information into the system.
Eight months later, the banker would receive a message generated by
the database to recontact this customer to discuss home loan options.
The Results
• During the first year, more than 1 million leads and $4 billion in
growth opportunities were sent to National bankers.
• During the next 6 months, 570,000 new leads were sent, which
resulted in the closing of $4.4 billion worth of new loans.
• During the second year of the system, premium sales of banking
products increased by 25 percent over the previous year, while
sales of wealth management products increased by 40 percent.
• The close rate for leads increased by five times over the close rate
before the new system began.
• The bank achieved a $391 million return on investment on one
campaign.
Quadrant Marketing
One valuable method of segmentation is by quadrants. The idea in
quadrant marketing is to take two variables, such as spending and
income, and divide all customers into four groups based on these variables. For example, let’s take a bank. Most banks and financial institutions
have learned to use a variety of inducements to capture the names and
transactions of all their customers. What they may not have is their customers’ incomes and ages. These can be obtained as appended information from Experian or other sources. Armed with these data, they
can develop four quadrants based on income and amount borrowed.
The resulting quadrants might look like Table 8-1.
175
176
Table 8-1 Quadrant Marketing
Q2 high income, low asset
Q1 high income, high asset
Population
% of total
Income
Marital
Average age
Average asset
Total assets
% of assets
Population
% of total
Income
Marital
Average age
Average asset
Total assets
% of assets
356,331
41%
$138,665
79%
38
$855
$304,663,005
5%
Q3 low income, low asset
Population
% of total
Income
Marital
Average age
Average asset
Total assets
% of assets
69,778
8%
$136,778
78%
52
$44,556
$3,109,028,568
51%
Q4 low income, high asset
377,221
44%
$55,669
53%
29
$766
$288,951,286
5%
Population
% of total
Income
Marital
Average age
Average asset
Total assets
% of assets
58,996
7%
$56,993
54%
61
$40,221
$2,372,878,116
39%
This bank has about 860,000 customers. Its total bank assets are
about $6 billion. They are divided into four quadrants: high income,
high bank assets; high income, low bank assets; low income, low bank
assets; low income, high bank assets
It is interesting that the high-income customers tend to be married,
whereas fewer of the lower-income customers are married. Also, the
people with high assets tend to be much older than the people with low
assets. We will set these facts aside for a moment and concentrate on
the incomes and assets.
Here’s how to do the math for a quadrant analysis. For income, look
at all household incomes. See if there are some at the top that are outrageous: $1 million, for example. Exclude the few outliers like this.
Then add all the incomes together and divide by the number of people.
For the data used to construct Table 8-1, the average income was
$68,230. Then divide customers into two groups: those with incomes
above and those with incomes below $68,230. Get the average income
for those above, which was about $136,000, and for those below, which
was about $56,000. Do the same thing with the average assets.
When you have done this, sit back and think about your table for a
day or two. What is it telling you? What are these segments, and what
can you do with them?
• Quadrant 1 marketing plan. Begin with the high highs. These people
have a lot of money, and a lot of it is in your bank. What is your
goal for these people? Retention. There is no point in marketing to
them. They are maxed out. All the money they have is in your bank,
and they can’t give you any more. But you don’t want to lose these
people. They are your Gold customers. So you design a retention
program for them and put them in it. Give them special benefits
that come with being Gold customers. They shouldn’t have to do
anything. They should be able to just sit there and let the benefits
roll in. Thank-you letters from the branch manager. Birthday cards.
Greetings whenever they come into the branch—each employee is
taught to recognize them and use their name. The branch manager
jumps up from his desk and shakes their hand whenever they come
in. If he sees them waiting in a long line, he goes up to them and
asks them to come to his desk, where he handles their transaction
personally. I have enjoyed that treatment at my bank, and I love it. It
works. I would never go anywhere else.
• Quadrant 2 marketing plan. These are the high lows. They have a
lot of money, but your bank has very little of it. They probably
have their money somewhere else. Here is an opportunity. Marketing is going to work here, because these people have assets
that they can manage, with your help. You might tell them the
benefits that come with Gold status at the bank, and show them
how easy it would be to reach Gold status by shifting deposits,
buying mutual funds, or engaging in other activities that would
benefit them.
• Quadrant 3 marketing plan. These are the low lows. Here is a group
of really worthless people, as far as you are concerned. They have
very little money, and they don’t put much of it in your bank.
Marketing to them would be a waste of your money, since they
don’t have much to work with. Here is a wonderful opportunity to
reduce your marketing expenses. These are 44 percent of the
bank’s customers. Spend your money somewhere else. You will have
money left over to concentrate on the other quadrants.
• Quadrant 4 marketing plan. Here is a really interesting group:
older people with a lot of money in your bank, but relatively low
incomes. Who might these people be? They are mostly retired,
of course. They have worked all their lives and they have saved
177
178
(or they inherited money), and they have something to show for
it. So what should you do? Well, in the first place, you don’t
want to lose these valuable people, so you try to get inside their
heads and think as they do. Focus groups would help here.
Many of them will need help with their wills. This is a free
service that you can provide for them. The bank might be the
executor if they have no one else. Most of them are not
interested in getting rich fast; they just want to keep what they
have. Long-term care insurance would be an obvious offering if
you handle insurance. If they are in anything risky, talk to them
about moving into something more stable, so that you don’t lose
them in a downturn. Become their financial adviser.
These may not be the best quadrant plans in the world, but they illustrate the power of this technique. They show what you can do with your
customer base to make it come alive and to let you direct your efforts
to areas where you can make more money for your company and make
your customers happy at the same time.
Reaching Mothers-to-Be
Kmart ran into some financial troubles in 2002, but that does not mean
that we cannot learn a great deal from some of Kmart’s excellent marketing programs. One of these programs focused on prospective
mothers. There is a lot of money to be made from selling to this group.
Mothers-to-be spend $18 billion annually, and 80 percent of that
amount is spent before the baby is born. For a 2-year period families
spend thousands of dollars in a category in which they have not recently
been spending and in which they are unlikely to spend again until the
next child arrives. Kmart and other retailers have understood the power
of this segment for some time and have made major efforts to capitalize on it.
Kmart developed a comprehensive strategy for targeting this segment.
Using its database, it was able to reach one-third of all U.S. moms-tobe before the baby was born. The basic strategy was this:
• Use the database to identify moms-to-be through their purchase
of maternity apparel and other prebirth items.
• Be the first to market before the baby is born, through databasetargeted direct-mail communications.
• Integrate with other departments as a cross-functional team to
change merchandise assortments and the arrangement of
merchandise in the stores so that the items purchased by mothersto-be are within easy reach.
Actual shopping behavior plus multiple regressions were used to determine which customers really were mothers-to-be (as opposed to
grandmothers or friends shopping for a baby shower). The data explored
the cross shopping of each individual, her market basket, her known age,
and her previous shopping history.
Armed with this information about the moms-to-be, Kmart developed test groups that got the special promotions and control groups
that did not. Comparison of the purchasing behavior of the two
groups clearly showed the advantage of the special targeting.
The Results
The program increased store sales by over 15 percent while reducing
mailing costs by 35 percent. In Hispanic markets, store sales of baby
products increased by over 25 percent, while sales in adjacent highly
cross-shopped departments increased by 35 percent over the prior year.
On a customer level, over 21 percent of the top-decile moms-to-be
shopped at Kmart during the mail event, increasing their spending more
than 40 percent over the prior year. Targeted customers outperformed
the control groups by 4 to 1 and spent over 20 percent more per visit
during the promotional event.
As a result of this experiment, the use of the database to drive promotions was expanded into many other departments at Kmart. According
to Tom Lemke, vice president of database marketing at Kmart, “Our
mission is to ‘Mail Less and Sell More.’ To accomplish this, we use the
intelligence of customer data to create improved response rates and
lowered mail volumes.”
179
180
How Many Segments Should You Create?
Everyone who tries to create customer segments runs into the problem
of numbers. By using recency, frequency, monetary (RFM) analysis, it
is possible to come up with hundreds of different RFM cells. By using
SIC codes, you can create hundreds of types of customers that you can
market to. By using profitability, you can subdivide your customers into
deciles, for example, from the most profitable to the least profitable.
How many segments make up the right number?
To get the answer, you need to consider the imagination and creative
genius available in your marketing organization. Every new marketing
idea requires time and resources for its execution. Few organizations
can develop and support as many as five or six simultaneous marketing
efforts. Therefore, the sector marketing strategy must end up creating
and marketing to just a few sectors.
Are Your Segments Viable?
OK. You have divided your customer base into five segments, and you
have developed a marketing program for each of the five. Let’s say you
divided your customers by level of total spending (rather than by age,
income, product categories, etc.). Were your segments the most profitable that you could have devised? How can you know?
One good way to find out is to create a universal control promotion
that you use with your segment control groups. This is a good promotion that has worked in the past and resulted in a lot of sales. It is used
to test the marketing promotions designed for each segment. Table
8-2 shows a possible result of a mailing to your segments based on total
spending.
Here’s how we did the math for the lift in sales. We subtracted the sales
rate for the control group from the sales rate from the test. In the Gold
segment, for example, we subtracted 7.94 from 8.10 and got 0.16
percent. Then we multiplied that 0.16 percent by the number of people in the test segment—in this case 100,000—to get the lift in sales.
What this tells us is that the creative material we used with the Gold
Table 8-2 Testing Segment Promotions
Customer
segments
Number
promoted
Sales
rate
Gold
Control
Silver
Control
Bronze
Control
Lead
Control
Total
100,000
20,000
100,000
20,000
100,000
20,000
100,000
20,000
480,000
8.10%
7.94%
6.60%
6.41%
5.01%
4.60%
2.34%
2.40%
Lift
Lift in
sales
0.16%
160
0.19%
190
0.41%
410
–0.06%
(60)
700
people resulted in 160 more sales than we would have gotten if we had
sent them the control package.
Unfortunately, this segmentation scheme is a self-fulfilling prophecy.
The high spenders spent a lot. The low spenders spent a little. You
knew that when you set up your segments. You could not think up
enough creative things to say to the Gold customers, for example, to
differentiate their behavior from that of the Gold customers who got
the same message that was sent to everyone else. This is bad segmentation strategy.
Suppose we were to create our segments in a different way. We might
do it by the age of the customer. We create special promotions for
people in their teens, people in their twenties and thirties, people in
their forties and fifties, and senior citizens. We want to see if we can
design messages that reach these segments and do better at selling to
them than a generic message. Table 8-3 shows a possible result.
As you can see, we have the same number of people, but they are
divided into age groups. The lift from our promotions is very much
more pronounced. The teen message really scored well and produced a
4025-customer lift over the control groups. We can’t say the same for
the work that our creative staff did for people in their forties and fifties.
From this you can see that the way you create your segments often
determines your results.
You should experiment with several segmentation schemes before
you settle on the one that you will use for a long time.
181
182
Table 8-3 Age Segment Promotion
Teens
Control
20s and 30s
Control
40s and 50s
Control
Seniors
Control
Total
Number
promoted
Sales
rate
115,000
20,000
145,000
20,000
75,000
20,000
65,000
20,000
480,000
9.50%
6.00%
7.70%
5.80%
5.00%
5.70%
4.00%
3.10%
Lift
Lift in
sales
3.50%
4,025
1.90%
2,755
–0.70%
(525)
0.90%
585
6,840
Rules for Segment Marketing
Based on this discussion, how do you go about profitable segment marketing? There are several rules:
• Build a customer database with demographics and transaction
history. Find a reliable way to get regular updates on transactions
and responses.
• Come up with a profitable method for creating customer segments
that you can market to.
• Select just a few segments and design a marketing program for
each segment.
• Set up control groups that have exactly the same characteristics as
the groups selected for marketing, but that do not get the
promotional material. Measure the performance of the segment
that received the promotion against that of the control group.
• In the marketing campaign for each segment, use the database
to create personalized content: emails, direct-mail pieces,
telephone calls, Web response pages, and so on. Monitor the
response from each segment and each response method and
record them in the database.
• Constantly test new methods of marketing to each segment,
recording the results and increasing the resources devoted to those
methods that work.
Summary of Case Study Results
• From iDine, we learned that you can track consumer behavior
by using registered credit cards. It is easy to do. You can
get many people to participate if you make it worth
their while.
• From the National Australian Bank, we learned that Teradata
computers and software permit you to “trawl” customer
accounts every night to find people who have had significant
changes in their bank accounts. We use segment marketing to
develop rules to market to each situation.
• From Kmart, we learned that you can identify mothers-to-be
based on their shopping behavior. Marketing to this segment was
very profitable for Kmart, and was worth the effort.
• From the University of Toronto, we learned that you can
reactivate lost customers using a creative database marketing
strategy.
What Works
• Dividing your customer database into a few segments based on
some logical scheme and developing a marketing strategy for
each segment.
• Always setting aside a control group in each segment that gets a
generic marketing promotion (or none at all) so that you can
measure the effectiveness of the promotions.
• Differentiating the promotions to each segment by personal data
in the database. “Since you bought the ski outfit last October,
you may want to consider a flannel nightgown to wear on those
cold evenings.”
• Finding a way to collect the data on what happened after you sent
the promotions. If you can’t get the data, you may be wasting your
money on the promotions.
183
184
What Doesn’t Work
• Creating more than eight or nine segments. Even the largest
companies don’t have the marketing staff to think up and implement
unique marketing programs for lots of different segments. If you
have divided your database into 10 or more segments, you have made
a serious mistake.
• Forgetting to create and use control groups. An advertising agency
created a “frequent user” segment with special benefits for a client.
The client spent more than a million dollars promoting to the
segment, and it seemed to work. After 2 years, the client asked the
agency to provide the value the client had received for the million
dollars. Unfortunately, there was no control group, so the agency
could not provide this information. The agency lost the account.
• Not revising your segments after each promotion. We are
supposed to learn something from the segments. If you just
promote and promote, and never revise, you may be wasting your
money and not learning anything.
• Collecting a lot of data about customers that you never use for anything. Everything has a cost. Getting and updating data is expensive. Every time you add a piece of information to your database,
ask, “How are we going to use this to make money? How can we
use this right away to modify customer behavior?” If you haven’t got
a good answer to these questions, don’t collect and store the data.
• Spending too much on building a data warehouse that will never
pay its way in terms of increased sales through promotions based
on the data in the warehouse.
• Creating segments that are self-fulfilling or that do not allow your
creative team to create a significant differentiation.
Quiz
1. Why do most marketers choose segment marketing over one-toone marketing?
a. One-to-one marketing does not work.
b.
c.
d.
e.
2.
3.
4.
5.
6.
Segments are easy to create.
They don’t know how to do one-to-one marketing.
Segments cost less but produce similar results.
One-to-one marketing produces a much higher lift
in sales.
Which of the following are not considered demographic data?
a. Age
b. Preferences
c. Ethnicity
d. Income
e. Value of home
What technique does iDine use that most marketers are not
using today?
a. Creating customer segments
b. Credit card matching
c. Air miles as a reward
d. Cash as a reward
e. Transactions viewed on the Web
Which is not listed as a feature of the Australian National
Leads program?
a. Web customer response
b. Customer segmentation
c. Nightly trawling of the data warehouse
d. Prediction of customer propensity to purchase
e. Offers based on transactions
What is the best strategy for Quadrant 3 – the low lows?
a. Retention
b. Expansion
c. Ignoring them
d. Repricing their services
e. Rewards for good behavior
Which was not a result of the Kmart mothers-to-be program?
a. Hispanic baby product sales were up 25 percent.
b. Cross shopping was reduced by 35 percent.
c. Twenty-one percent of top-decile shoppers increased their
spending by 40 percent.
185
186
d. Targeted customers outperformed the control by 4 to 1.
e. None of the above; all were results of the promotion.
7. Compute the lift in sales in the following table.
Teens
Control
20s and 30s
Control
40s and 50s
Control
Seniors
Control
Total
Number
promoted
Sales
rate
100,000
20,000
100,000
20,000
90,000
20,000
60,000
20,000
430,000
8.80%
7.00%
6.60%
5.80%
7.20%
6.00%
5.10%
4.30%
Lift
Lift in
sales
8. Which of the following is the best number of segments to create
for segment marketing?
a. 100
b. 80
c. 50
d. 12
e. 6
9. What is the best way to design segments?
a. Something that you can relate to and design programs for
b. RFM groupings
c. LTV segments
d. SIC segments
10. When should segments be revised?
a. Monthly
b. Quarterly
c. Never; keep them consistent
d. After each promotion
e. Once a year
C
H A P T E R
9
Helping Business
Customers to
Become Profitable
M
ost manufacturers sell their products through wholesalers, dealers, brokers, or agents. They don’t have to deal directly with the
ultimate consumer of the product. This situation has both a good and
a bad side. The good side is that the manufacturer is selling in wholesale quantities. It has hundreds instead of millions of customers to worry
about. The bad side is that if the intermediaries are not good at their
business, sales will falter, and there is not much that the manufacturer
can do about it. It is hard to introduce new products when you cannot
communicate with your ultimate consumers.
Database marketing has provided a new method of handling the marketing of business products. If the manufacturer can just get its hands
on the names and addresses of the ultimate consumers, it can correspond with them, introduce them to new products, and drive them to
the dealers. Many dealers resist this approach. Their fear is that the
manufacturer will use this information to go around them and sell
products directly to the ultimate consumers.
This situation almost drove Compaq out of business. Compaq had a
network of wholesalers and retailers that were the exclusive marketers
of its computers. Beginning in the middle 1980s, Compaq developed an
excellent database of registered users of Compaq products. The company knew their names, their addresses, their positions in business, their
occupations, the software they used, and scores of other pieces of information about them. But it never once contacted any of them directly,
for fear that its retailers would stop promoting the Compaq name.
187
188
Meanwhile Dell, Gateway, and then IBM began to sell PCs directly to
the public, stealing market share away from Compaq. When Compaq
finally woke up to its losing strategy and began to sell direct, it was too
late. Compaq had dropped to the back of the pack.
With database marketing, the ideal strategy for a supplier is to work
to make its dealers, agents, and brokers successful. The supplier wants
to help them to succeed in their business. As a result, the dealers will
push the sale of the supplier’s products. Presented in this chapter are
several case studies in which suppliers have done just that.
Top 10 Prospects
A manufacturer of industrial components made 40,000 different products and had $400 million in sales through 1800 independent agents.
While $400 million may sound like a lot, it was less than 1 percent of
the $87 billion in annual sales in this business category. The independent agents handled the products of many other manufacturers in addition to this firm’s. Some agents were very good at pushing this firm’s
brand. Others were not so good. The company wanted to stand out
from the crowd. It turned to Scott Brostoff of CSC Advanced Database Solutions to come up with a method of helping its agents succeed.
Each agent had a unique territory. The firm had compiled a list of hundreds of thousands of ultimate customers based on their deliveries over
3 years. CSC decided to take this database and add Dun & Bradstreet
data to it, showing for each customer the SIC code, annual sales, number
of employees, and actual products that the customer had purchased.
What types of data are available from Dun & Bradstreet for
appending to a business-to-business file? Some different types are
listed in Figure 9-1.
Scott decided to feed the D&B information into a model that showed
which products in which quantities customers in each SIC code bought.
Armed with that information, CSC decided to use the D&B data to identify prospects in each agent’s territory that were not currently buying the
firm’s products. The model was designed to score each prospect in order
to determine the 100 prospects in each agent’s territory that would be
H e l p i n g B u s i n e s s C u s t o m e r s t o B e c o m e P r o f i ta b l e
Figure 9-1
Data
Dun & Bradstreet
Dun & Bradstreet Appended
Business Data
Company name and address
Business family linkage
SIC codes
Line of business
Territory covered
Number of employees
Trends
Year business started
Bank relationships
Contact names
Net worth
Annual sales
Financial stress scores
Energy demand: electricity, gas
Telecom demand
Number of PCs
IT spending
Propensity to lease
Market penetration
most likely to consume the largest amount of the firm’s products. The
resulting prospect database of 180,000 companies was designed to list
the top 10 prospects in each agent’s territory and the particular products
that these prospects were most likely to buy. The model was also
designed to predict the next best product for each of the existing customers in each agent’s territory. The next best product was the product
that the customer was not now purchasing that the model predicted was
most likely to be purchased and to have the highest sales.
This was information that the agents did not have and could not easily
develop without CSC’s help. CSC’s next job was to get the agents to
buy into the idea and subscribe to the top 10 prospects and next best
product information. Experience in other situations had convinced CSC
that agents will use valuable information like this only if they pay for it.
If it is offered free, they will consider it worthless. How do you get 1800
independent agents to sign up for such a service and pay for it?
Scott decided to develop a major campaign to sell the agents. He
started with the few agents who he knew would buy into the idea from
189
190
the start. Their success in using the top 10 prospects could be used to
convince the others that they could profit by signing on.
The delivery system for the names was interesting. CSC decided to
install Epiphany software on the database which could be accessed by
the 1800 agents through the Web. Each participating agent received an
ID and a PIN so that its employees could sign onto the database and
see a Web page that was designed just for them and that featured
information about their sales and commissions, their top 10 prospects,
the next best product for each of their existing customers, and a chart
that compared their sales with those of other agents in their region and
in the nation. A contest was designed for the agents who made the most
effective use of the top 10 prospects, with prizes for the winners. By
making its independent agents profitable, the manufacturer would
increase its sales and profits.
How to Determine the Next Best Product
One of the most practical and useful applications of database marketing for a business-to-business company is determining its customers’
next best product (NBP). The NBP is a guide for the sales force. Typically, the sales force tries to get companies to buy more of what they
are already buying. That makes sense, and it is logical and easy to do.
What many salespeople have found, however, is that their best customers are “maxed out” in certain categories. They are already placing all
their orders for a category with this one supplier, and so pressures to
buy more may not be effective. So what else do you sell to this company? Suppose you have 40,000 different products in your catalog.
Which ones do you push? Until the advent of database marketing, the
answer depended on intuition, observation, and luck. To these three,
we can now add one more: next best product analysis. Next best product analysis will tell you what to sell to whom. It divides customers into
similar buying groups, examines what each group is buying now, and
then scores each customer on the basis of the relationship between sales
to the customer and sales to other members of its group. It can be a
very powerful guide for the sales force.
For a business-to-business application, how do you go about determining the next best product for each of your customers? Let us assume
that you already have a database with information on your customers
and their purchases for the past 2 years and that you have overlaid your
file with SIC codes, number of employees, and annual sales. These data
can be obtained from D&B.
Your first job is coming up with meaningful SIC categories. For each
SIC group, determine your total sales. You may have to do some grouping to get something to work with. For example, Table 9-1 gives just a
few out of perhaps a hundred SICs.
I would not waste time on SICs B and F. They should be combined
with others. You might want to collapse SIC E as well. You will end up
with perhaps 40 meaningful SICs or even fewer.
Then I would divide the final SICs into categories on the basis of
either number of employees or annual sales, depending on your preference. We can divide them into three categories as shown in Table 9-2.
We now have divided our customer base into 21 different groups on
the basis of SIC group and size.
Table 9-1 Sales by SIC Groups
SIC
A
B
C
D
E
F
G
Your sales
$1,889,000
$23,000
$4,599,000
$102,000
$56,000
$124
$3,445,220
Number of
companies
14
1
18
3
15
1
178
Average
sales
$134,929
$23,000
$255,500
$34,000
$3,733
$124
$19,355
Table 9-2 Customer Companies by Number of Employees
Number of employees
SIC
Your sales
Number of
companies
Under 400
401–1000
1001+
A&B
C
D, E, F & G
H
I
J
Total
$ 1,912,000
$ 4,599,000
$ 3,603,344
$ 1,002,000
$ 4,567,000
$ 2,089,000
$17,772,344
15
18
197
41
101
43
415
$ 552,001
$ 45,000
$2,001,000
$ 200,000
$1,002,000
$ 45,001
$3,845,002
$ 200,000
$ 456,000
$ 500,000
$
62,000
$ 890,000
$ 1,998,000
$ 4,106,000
$1,159,999
$4,098,000
$1,102,344
$ 740,000
$2,675,000
$ 45,999
$9,821,342
191
192
What Do They Buy?
Next, we turn our attention to the products sold to these companies.
We have to do some consolidation here as well. Some companies are
buying 100,000 different products, and others are buying only 4 or 5.
To make sense out of these numbers, we have to break these products
down into meaningful categories. By meaningful, I mean that not only
do they have to have sufficient sales to be worth bothering with, but
they also have to have some relevance to one another. For example, we
might have categories like those shown in Table 9-3.
Looking at these categories, I might concentrate on the four product
categories in which the average sale per buyer is $27,000 or more. Depending on the situation, you might also combine some of the categories.
Now it is time to use a sophisticated model. We take each of our 415
current customers and determine that customer’s sales in each of our major
product groups. We look at the SIC group and the employee size category. The model will create a next best product for each of the 415
different customer companies based on that company’s product purchases
relative to those of the other companies in its SIC and company size group.
Each of the 415 different company scorecards will look like Table 9-4.
Table 9-3 Average Sales by Product Type
Hoses
Fasteners
Valves
Pipe
Wire
Switches
Connectors
Total
Total
sales
Total
buyers
Average
per buyer
$ 6,001,000
$ 778,001
$ 1,002,000
$ 9,001,000
$ 339,000
$ 603,000
$
46,001
$17,770,002
220
198
18
144
201
22
32
415
$27,277
$ 3,929
$55,667
$62,507
$ 1,687
$27,409
$ 1,438
$42,819
Table 9-4 Next Best Product for XYZ Company
Total average annual sales
SIC group average
This company
Next best product
Hoses
Valves
Pipe
Switches
Total
$38,002
$22,001
$60,003
$78,003
$42,001
$ 1,001
$42,001
$2,002
$8,990
$142,008
$109,995
What Table 9-4 says to the sales force is that the next best product
to sell to XYZ company is pipe and that the annual sales could amount
to $42,000 per year, based on the purchases made by other similar companies.
These numbers can be produced for every company and given to the
sales force. Each salesperson can be measured not only on total sales
but on the ability to increase sales in these conquest categories.
Computing the 10 Best Prospects
Suppose that you are a business-to-business manufacturer with a sales
force, each member of which has an assigned territory. Some of the
salespeople are doing well, and some are not doing so well. The sales
force may be company-owned, or it may consist of independent agents.
In either case, you want to equip your salespeople with the most productive data to help them in their efforts. One vital sales assist is the
NBP described previously.
Even more powerful can be a top 10 prospects list developed for each
unique sales territory. Here is how such a list is developed. A part of the
next best product determination was the creation of SIC categories,
broken down by small, medium, and large companies. Using that analysis, you can rank each of your existing customers by annual sales. This
might produce a ranking for a single territory that looks like Table 9-5.
For the top 10 analysis, we may arbitrarily confine our attention to
companies that have annual sales of over $100,000. There are three such
Table 9-5 Ranking of
Sales within a Territory
Company
Sales
G
B
C
D
E
A
F
Total
$1,992,001
$ 889,445
$ 102,004
$ 56,002
$
1,245
$
317
$
77
$3,041,091
193
194
companies in Table 9-5. Because of the NBP analysis, we know the product that we sell the most of to each of these three companies and the companies’ total annual purchases from us, and we have categorized them as
being large, medium, or small companies. We also know what categories
of products the top three companies in this territory buy (see Table 9-6).
Now let’s go to D&B to find all the companies (that are not current
customers) in this particular sales territory that match the profiles of
companies G, B, and C (the companies with over $100,000 in annual
purchases from us). We may find that there are 34 such companies.
Table 9-7 indicates what the analysis will show us.
Prospects that resemble company G are clearly the best prospects for
future sales. And we know what to offer them: hoses. That is a nobrainer. Where do we come up with the next four to round out our top
ten? We have to pick, arbitrarily, four of the eight companies that are
similar to company B. We can offer them switches.
The way top 10 prospects works, however, we are not yet through
with our analysis. In a typical process, each agent is linked to the central
database by a Web interface. When an agent logs on to the database, he
sees a list of all his customers, together with the next best product for
Table 9-6 What the Top Three Customers Buy
Total
sales
Hoses
Fasteners
Valves
Pipe
Wire
Switches
Connectors
Total
$ 6,001,000
$ 778,001
$ 1,002,000
$ 9,001,000
$ 339,000
$ 603,000
$
46,001
$17,770,002
Total Average
buyers per buyer
220
198
18
144
201
22
32
415
$27,277
$ 3,929
$55,667
$62,507
$ 1,687
$27,409
$ 1,438
$42,819
Company
G
Company
B
Company
C
$ 889,554
$ 19,646
$ 478,003
$ 312,535
$
8,433
$ 276,881
$
7,188
$1,992,240
$227,665
$ 3,929
$ 55,667
$ 62,507
$ 1,687
$545,667
$ 2,434
$899,555
$ 19,882
$ 1,965
$ 27,833
$ 37,253
$
843
$ 13,705
$
719
$102,200
Table 9-7 Top Products of Prospect Companies
Company
G
B
C
Total
Similar
Sales
companies
$1,992,001
6
$ 889,445
8
$ 102,004
20
34
Best-selling
Average
product
annual buys
Hoses
$889,554
Switches
$545,667
Pipe
$ 37,253
each company. He also sees a list of the 10 best prospects, along with
the product that he should sell to them. By clicking on each prospect,
he gets the name and address of the company, the contact information,
company data, and so on.
Let us suppose that this particular agent has already canvassed some
of these prospects and knows that he cannot win them over. They are
already locked into the competition for some solid reason. The agent
needs more prospects. No problem; he clicks on a “Do Not List” column opposite each of these prospects, and then clicks on a “Refresh”
button. The top 10 prospect software goes back to the prospect pool
and redoes the analysis, dropping the “Do Not List” companies and
adding new companies from the pool. Within a few minutes, the agent
has a refreshed list of top 10 prospects. He can do this as often as he
wishes, working his way down all the prospects reported by D&B
within his territory.
This is an example of the way modern database analysis supports
business-to-business marketing and sales. The information shown in
these tables could not have been developed without a customer marketing database. Once you have that database, the software to produce
next best product and top 10 prospects is complex, but achievable.
Using the Web Correctly
BenefitMall.com is an online exchange for buyers, sellers, and providers
of employee benefits for companies with 100 employees or less.
Through BenefitMall, insurance brokers can access active accounts and
receive accurate real-time quotes, reducing the total sales cycle from
more than a month to less than an hour in most cases. As well as the
Internet, BenefitMall also supports brokers with traditional in-market
call center and sales support. The Mall’s quote engine lets brokers quote
multiple carriers and multiple products in a single proposal.
There is a large community of brokers in the United States that focuses
on providing insurance and benefits contracts to small businesses. Also,
there are many companies that offer small-company benefit packages, with
many different product offerings. The challenge for a broker is to be able
195
196
to access information on a variety of options in a timely fashion. Since brokers usually have small budgets and cannot afford sophisticated systems,
they find companies like BenefitMall to be quite helpful. Going directly
to a carrier can require a lot of paperwork and take weeks to process. By
the time it’s all completed, the client may have chosen to purchase from
some other broker.
BenefitMall’s director of marketing, Patty Harris, asked SmartDM
Chief Analytics Officer Bill Franks to help the company make a major
expansion of both its online and its offline capabilities and staff. With more
than 50,000 brokers in the company’s database, they decided to launch a
program to increase the share of business obtained from brokers.
The goals of the initiative were to
• Set up a new Web-based database marketing system across several
channels, including direct mail, email, fax, and Web surveys.
• Provide regular updates to broker segment classifications and
respond rapidly to changes in the broker profile.
• Begin marketing to brokers through programs based on the type
and frequency of business that they have done with BenefitMall.
• Allow easy collection of broker information, including
communication preferences, demographics, and customer
satisfaction feedback.
• Have responses tracked and results quantified so that the
effectiveness of marketing efforts can be rapidly determined and
the efforts adjusted as needed.
• Personalize broker communications with the data collected and the
name of the broker assigned to each broker.
To set up the new system, there were several obstacles to be overcome:
• The BenefitMall database was not “marketing-friendly.” Getting
data out of it was difficult.
• BenefitMall had not done sophisticated database marketing in the
past. It was difficult to sell the concept to management.
• There was no historical information that could be used to
benchmark the response to new programs.
• Management, once it was sold on the program, demanded an
aggressive implementation process, with not enough time for testing.
Program Communications
To set up the new system, Franks and the SmartDM team set up a number of new communications aimed at building relationships with brokers:
• Welcome kit mailings to new brokers when they first register on
the system.
• Activation emails encouraging brokers to make their first quotes
and sales.
• New customer thank-you mailings to thank brokers for their first
BenefitMall sale.
• Thank-you satisfaction surveys emailed to each broker after each
sale. The email links the broker to a Web microsite that asked
questions about the experience. The emails came from the
regional vice president of sales for the broker’s region.
• Reactivation emails, letters, and phone calls of two types: one for
those who stopped selling, and one for those who stopped quoting.
• Broker newsletter emails: nine regional monthly versions.
• Special regional postcard/email/fax promotions. These were
special communications aimed at getting information such as new
carrier announcements, rate changes, and so on to brokers in a
short period of time.
Database Changes
For its marketing database, BenefitMall used SmartDM’s DirectTR@K
system, which got updates every 2 weeks from the BenefitMall system.
This database stored all the information about each broker: past transactions, past marketing communications received, when the broker
became a customer, and so on. The DirectTR@K system sent the emails
automatically. It tracked who opened the emails, who clicked on the
various links, and what emails bounced. While the DirectTR@K system
sent the email communications directly, lists for the direct-mail and fax
197
198
communications were extracted and sent to SmartDM’s production
division for execution.
The communications were personalized and customized to include
transaction data, sign-up date, and the sales rep and sales manager and
their contact information. In this way, the communications appeared to
have been sent to the broker by the rep whom the broker knew.
New Systems
This program was entirely new for BenefitMall. It required setting up
• A Web-based, multichannel, marketing system that tracked
customer activity and response.
• Data of much higher integrity than had previously been maintained.
It was necessary to pull lists on the same day as the request, which
was very new to BenefitMall.
• User interfaces for generating queries, examining data on individual
brokers, managing campaigns, and viewing marketing reports.
• Broker profiles, including type of practice, product sales, and
transaction types. This information had not been available before.
• Customer satisfaction data from the Web surveys, which enabled
BenefitMall to identify strong and weak personnel within the
company.
• One-on-one direct marketing communications across several
channels (email, direct mail, fax, phone, and Web microsite).
With the new program, the company added a BenefitMall custom
proposal, a Microsoft Excel–based software tool designed to make it
easier for brokers to customize insurance coverage proposals for clients.
The goal of this free software was to help brokers customize proposals
so that they could sell more insurance. The Web system (see Figure 9-2)
permitted brokers to
• View side-by-side rate and benefits comparisons for multiple policies.
• Add coverage options to help clients compare rates and benefits.
Figure 9-2
BenefitMall Web Site
• View preloaded, preset spreadsheets and reports.
• Create final presentations right on their laptops, then view the
proposal on the screen or print it out and discuss it.
After a 120-day development process, BenefitMall was able to start
its first month of the new system by sending
• Welcome kits to 2500 new brokers, personalized for the individual
broker and the broker sales representative for each of the 11 regions.
• Activation emails to all new quoting brokers and all new selling
brokers. BenefitMall was able to convert 5 to 10 percent of these
brokers to full customers within the first 2 months. This was true
even though typical sales cycles are 3 to 6 months. It is anticipated
that this percentage will at least double over time.
• New customer thank-you letters and gifts to all first-time selling
customers in all regions.
• Thank-you satisfaction surveys to all brokers completing sales
each month; 25 percent of the brokers reached in the emails
participated in the surveys.
• Reactivation emails, letters, and sales phone calls to all lapsed
brokers, those who had stopped quoting and selling. BenefitMall
199
200
was able to reactivate more than 5 percent of these brokers within
the first 60 days. This percentage is expected to at least double,
since typical sales cycles are 3 to 6 months.
• Email newsletters to all brokers with 45 to 55 percent open rates
for HTML. (Open rates refers to the percentage of people who
open and read an email instead of just deleting it.) There was a 5
percent click-through rate on links to the Web site and an opt-out
rate of 0.3 percent over an 8-month period.
• Eighty special regional postcards, faxes, and emails, all within 48
hours of the event, improving the turnaround time by 62 percent.
These communications gave brokers updates on price changes,
seminars, new carriers, or other time-sensitive news items.
In the first 8 weeks of the new program, BenefitMall generated over
2000 new quotes and 80 new sales.
Postcard Direct
Isuzu dealers sell midsized trucks that are used in a variety of occupations, including landscaping, plumbing, manufacturing, and materials
distribution. Because of the differing vocations for which the trucks are
used, the engines are the same but the appearance of the trucks varies
considerably. To sell a truck to a nursery or an electrical contractor, it
is useful to show the prospect a picture of what the truck will look like.
To solve this problem, Isuzu worked with msdbm of Los Angeles to create a program called Postcard Direct.
Isuzu asked its dealers to provide it with scores of testimonials from
satisfied Isuzu truck owners, complete with photographs of the trucks
being used in various vocations. Msdbm helped Isuzu to assemble the
best of these testimonials and photographs into a group of oversized
postcards designed to be sent to customers in various vocations. Msdbm
worked with Dun & Bradstreet and Isuzu to compile a database of small
and midsized businesses and prospects, coded by business type. These
names were put into a database with the latitude and longitude recorded
in each record. In addition, the database contained the names, addresses,
phone numbers, and logos of all the Isuzu dealers in the United States.
To launch the program, msdbm created a mailing to all Isuzu dealers,
inviting them to come to an Isuzu microsite (see Figure 9-3) to order
custom postcards to be sent to customers and prospects in their trade
area. Dealers could come to the site and select the vocations that they
wanted to promote, select the number of names they wanted to mail to,
and view the test front and back of each postcard.
After dealers selected the postcards they wanted to send, they would
see the number of prospects and customers whose offices were within
a close radius of the dealership (see Figure 9-4). They could order the
cards right on the Web. Dealers would click on the number of prospects
they wanted to mail to and then would see a form for placing the order
(see Figure 9-5).
The program proved to be an outstanding success. Fifty percent of
the Isuzu dealers in the United States logged on to the Web site and
ordered postcards. After 1 year, a total of 128,000 postcards had been
ordered using the system, at an average cost to the dealer of $0.25 each.
Figure 9-3
Postcard Direct
201
202
Figure 9-4
Order Form for Postcard Direct
Figure 9-5
Postcard Order Details
Along with the postcards, the dealer got printed labels to affix to the
postcards. The dealer could mail the cards on whatever schedule was
suitable to keep the sales force busy.
This program was an extremely effective use of the Web. There was
no other way that such a program could have been run in such a costeffective manner. Direct mail would have been much more costly. Phone
would not have worked, because the dealers would not have been able to
view the postcards. Once the dealer ordered the cards, they were mailed
within a week, together with the labels. The dealer got an immediate
email thanking him or her for the order. The dealer was also able to get
an email list of the names and phone numbers of the prospects to which
the cards were to be sent so that the salespeople could follow up.
Secret Double Agents
Winning the hearts and minds of travel agents is one of the keys to running a successful hotel chain. Agents are barraged by faxes, brochures,
and advertisements from hotels, airlines, rental car companies, cruise
lines, and destinations. The information changes daily as rates are
revised and promotions generated. If a hotel wants to get the attention
of travel agents, it must break through this information overload and
promise something more.
To increase booking activity for Radisson Hotels during slow travel
periods, Gail Recker of the Carlson Marketing Group created a “hurdle”
promotion to motivate travel agents. Using the Radisson travel agent
database, Gail created a specific “confidential” and achievable goal for
each agent’s booking of Radisson hotels between March and May.
The goal varied based on an analysis of the agents’ past activity with
Radisson. Upon reaching the goal, the travel agents would be rewarded
with “double points” for bookings made through August. The points
earned by the travel agents would put them into a sweepstakes drawing.
Eight agents would be selected in the drawing for an all-expenses-paid,
2-week trip to one of several exotic spots as a training exercise.
Because the reward for agents offered double points, the program
was called Secret Double Agent, with spylike teasers and a mission for
each travel consultant.
203
204
Phase 2 of the promotion focused on the slow travel period just
before the holiday season, with successful agents being rewarded with
double points for future bookings made in October and November.
Radisson and Country Inns & Suites general managers received a mailing
to alert them to the programs.
How the Hurdles Were Determined
The program was launched by sending a self-mailer with customized
hurdles to 80,000 U.S. and Canadian travel agents. The hurdle levels
were determined by segmenting agents into 10 tiers based on past
booking patterns, probability, and liability. A control group of 20,000
agents was identified that was not included in the promotion. This gave
Carlson the ability to measure the incremental value of the program.
The Radisson database, which had been built up over 9 years,
included 250,000 travel agents worldwide. The database was used to
communicate monthly with the agents.
The goal of the program was to increase incremental revenue from the
80,000 targeted agents by between $1.5 and $3 million over a 6-month
period. The results were far greater than anticipated. The incremental
revenue increased by $4.5 million, with 15 percent of the targeted agents
reaching their hurdle and receiving the double points as an incentive.
Supply Chain Management
Probably the most important new use of database marketing and the
Web is in supply chain management. There are many different inputs
that manufacturers need in order to make their finished products. Keeping the warehouse full of parts entails high interest payments and
requires lots of space and employee management.
Now that everything in the world is bar-coded, it has been possible
to introduce vendor-managed inventory. The system works as follows:
A manufacturer keeps its inventory records on a computer that can be
accessed by its suppliers through the Web. It works out arrangements
with its suppliers so that they can use the Web to learn electronically
every hour what parts have been used and which remain on the manufacturer’s shelves. The manufacturer discloses its sales plan so that the
suppliers can estimate how many of which items will be needed each day.
Then they can arrange just-in-time shipments to keep the inventory to
an absolute minimum, but always sufficient so that the production line
is never halted for lack of a part. Before the Web, some manufacturers
used a system called EDI (Electronic Data Interchange), which served
some of the same functions as vendor-managed inventory. EDI, however, was much more difficult to install and operate than the Web.
Using the vendor-managed system allowed Dell to hold less than 8 days’
worth of inventory at any given time. Without the system, Compaq had
to maintain 3 weeks’ worth during the same period. This gave Dell a tremendous cost advantage. Intel shipped products to Dell three times a
week without Dell’s having to process a single request—Intel could forecast Dell’s needs based on Dell’s sales forecasts and current stock usage.
Let’s see how a supplier might profit from using vendor-managed
inventory and “forward-deployed” products for 400 of its 1000 best
customers. These customers spend much more than other customers—
that is why they are the best. Figure 9-6 shows what the new system
might look like if you were the supplier.
Your products will be stored in the customer’s warehouses and in
your nearby warehouses, with title still remaining with you. Every time
your customer takes your products from its shelves or from your warehouse, your vendor-managed inventory system will know this. Shipping
Figure 9-6
Vendor-Managed Inventory
Web real–time interface
Customer
inventory
system
Customer
site
Vendormanaged
inventory
Your
site
Shipping
Billing
205
206
will be notified if it is necessary to replenish products in inventory.
Billing will be notified to bill the customer. Using this system, the following things will happen:
• The customer will save money: It no longer has to manage its
warehouses, nor does it have to pay for any inventory until that
inventory is actually used. This is a very large dollar saving to the
customer.
• The retention rate for customers using this system will approach
100 percent. Customers’ spending on your products will also
substantially increase, since, if they like the system, they may
begin to shift to your company’s products instead of competitors’.
• Your costs will not increase. Of course, you have to maintain
inventory at remote sites, but you were doing that anyway.
• We assume that your customers already have an inventory system
that keeps track (using bar codes) of the disposition of each
product in the warehouse. You can use the Web to poll this system
and get hourly reports on the movement of inventory. You have to
add the vendor-managed inventory software, which
• Does the hourly polling of inventory status at each customer
site
• Automatically replenishes products when they run low and
recalls products when they are not moving
• Forecasts usage and automatically increases or decreases
minimum stock quantities
• Automatically bills the customer for products taken off the
shelves
• Provides complete reports for you and for all your
customers
Table 9-8 shows what your 400 best customers will look like before
you introduce vendor-managed inventory. What does VMI cost? That
depends on the complexity of your operation. Let’s assume that it costs
$1 million to set it up, which annualizes to $350,000 per year, as shown
in Table 9-9. What could happen to the 400 best customers who adopt
this system? See Table 9-10. The effect of the new system on total profit
Table 9-8 Base LTV
Year 1
Year 2
Year 3
Retention rate
Customers
Annual spending
Revenue
75%
400
$50,000
$20,000,000
80%
300
$52,000
$15,600,000
85%
240
$64,000
$15,360,000
Costs
Operating costs
Marketing cost ($50)
Web site costs
Total costs
70%
$14,000,000
$ 160,000
$
20,000
$ 270,000
$14,450,000
68%
$10,608,000
67%
$10,291,200
$
15,000
$ 270,000
$10,893,000
$
12,000
$ 270,000
$10,573,200
$5,550,000
1.10
$5,045,455
$5,045,455
$12,613.64
$4,707,000
1.19
$3,955,462
$9,000,917
$22,502.29
$4,786,800
1.28
$3,739,688
$12,740,604
$31,851.51
Profit
Discount rate
NPV of profit
Lifetime value
Table 9-9 Annual Costs of
VMI Software
Software development
Annualized
$1,000,000
$350,000
Table 9-10 Improved LTV
Year 1
Year 2
Year 3
Retention rate
Customers
Annual spending
Revenue
90%
400
$54,000
$21,600,000
93%
360
$56,000
$20,160,000
96%
335
$68,000
$22,766,400
Costs
Operating costs
Web site costs
VMI costs
Total costs
70%
$15,120,000
$ 160,000
$
20,000
$ 270,000
$ 350,000
$15,920,000
68%
$13,708,800
67%
$15,253,488
$
18,000
$ 270,000
$ 350,000
$14,346,800
$
16,740
$ 270,000
$ 350,000
$15,890,228
$5,680,000
1.10
$5,163,636
$5,163,636
$12,909.09
$5,813,200
1.19
$4,885,042
$10,048,678
$25,121.70
$6,876,172
1.28
$5,372,009
$15,420,688
$38,551.72
Profit
Discount rate
NPV of profit
Lifetime value
207
208
Table 9-11 Profit Gain with New System
LTV after
LTV before
Increase
With 400 customers
Year 1
Year 2
Year 3
$12,909.09
$12,613.64
$ 295.45
$ 118,182
$25,121.70
$22,502.29
$ 2,619.40
$1,047,762
$38,551.72
$31,851.51
$ 6,700.21
$2,680,084
for these top 400 customers is shown in Table 9-11. The result is a
profit of $2.6 million over 3 years.
Do these numbers prove that Web pages for your best customers and
vendor-managed inventory for your very best customers will work in
your case? Of course not. You will have to work out the numbers for
yourself. What this chapter shows you is how to go about it. Develop
tables like these for yourself to see if these systems will work. Then, if
the numbers look good, go ahead and see if you can
• Find the software you need at a price that will make the numbers
profitable.
• Sell your best customers on adopting such a system. This will be
the most difficult part of the job.
Panduit’s Vendor-Managed Inventory
One of the pioneers in vendor-managed inventory is Panduit, which
makes 60,000 different electrical products such as switches, fuse boxes,
connections, and so on. The company markets solely through 1800 distributors. For its largest customers, Panduit set up a fully automatic
computer-to-computer management system. The distributor and Panduit establish mutual goals for service level and inventory turns. The
distributors then transmit a daily report on changes in inventory. Without human intervention, the system automatically
• Forecasts demand for each part number monthly.
• Determines the optimum replenishment quantity.
• Creates purchase orders in Panduit’s and the distributor’s business
systems.
• Ships and transmits replacements and the associated electronic
paperwork.
The advantage of this system is that Panduit has solved its customer’s
business problems by having
•
•
•
•
•
•
Freight costs reduced.
Human errors eliminated.
Operating costs minimized.
Out-of-stock conditions eliminated.
Profits increased.
Distributor loyalty maximized—the distributors become
relationship buyers.
FedEx Shipping Partnerships
National Semiconductor, a major maker of computer chips, handed over
its logistics management to FedEx. Before the transfer, National
shipped products from its six factories to seven warehouses around the
world. FedEx set up a single warehouse in Singapore to house all of
National’s output. National’s customers place orders with National’s
order center in Santa Clara, California. Many of these orders are placed
directly on National’s Web site, www.national.com. When the orders
come in, they are processed electronically and sent to the FedEx shipment center in Memphis. There the orders are routed electronically to
the FedEx warehouse in Singapore, where the products are picked,
packed, and shipped by FedEx to National customers throughout the
world. The result of this system has been:
• A reduction in the customer delivery cycle from 4 weeks to
1 week
• A reduction in National’s distribution costs from 2.9 percent of
sales to 1.2 percent of sales
• Elimination of the seven National regional warehouses in the
United States, Asia, and Europe
209
210
The FedEx–National partnership enabled National to get a jump on
its competitors by building profitable relationships with its customers.
By combining vendor-managed inventory with FedEx shipment,
National could concentrate on its main jobs—designing, manufacturing, and selling chips—rather than storing and shipping the chips.
Summary
What do these case studies tell us about how database marketing and
the Web are helping suppliers to make their customers profitable?
• In the first case, CSC decided to analyze the firm’s customer
database, append D&B information, and determine the next best
product and top 10 prospects for 1800 independent agents.
• BenefitMall.com used the Web to help thousands of independent
brokers determine the best benefit systems for their customers,
signing up 2500 new brokers in the process.
• Isuzu used the Web to provide its dealers with customized
postcards to be mailed to prospects in their area.
• Radisson Hotels used the Web to encourage 80,000 independent
travel agents to increase bookings by $4.5 million, benefiting both
the agents and Radisson.
• Intel used the Web to conduct vendor-managed inventory for Dell,
thus reducing Dell’s inventory to 8 days, while Dell’s competitor
Compaq had an inventory level of 3 weeks.
• FedEx used the Web to help National Semiconductor cut the time
required for delivery of parts to customers by 3 weeks per order
and cut delivery costs in half.
None of these activities would have been possible without a customer database or without the Web. The combination of the two has
created a powerful shift in the relationships between suppliers and
their customers.
What Works
• Vendor-managed inventory over the Web that permits suppliers to
manage their customer’s inventories.
• Providing agents with specific next best products for each
customer and top 10 prospects.
• Finding out more about the customer’s business and solving the
customer’s business problems.
• Creating a database of prospects and serving them up to agents
via the Web.
• Taking over logistics management for customers, so that they
can concentrate on their core business: designing and
manufacturing products.
What Doesn’t Work
• Old-fashioned business-to-business ordering systems using fax and
phone to order products that are used daily and in quantities.
• Manufacturers not concerning themselves with the business
problems of their agents.
• Assuming that intermediaries can do their job without your help.
Maybe they can, but why take a chance?
Quiz
1. In the CSC business-to-business case, how were the top 10
prospects for each agent selected?
a. Agents were asked to select them.
b. Existing customers were modeled.
c. Prospects were surveyed.
d. Prospect sales were modeled.
e. D&B provided prospect purchase intentions.
211
212
2. Which of the following is not typically part of D&B businessto-business data?
a. Market penetration
b. Number of employees
c. Prizm cluster codes
d. Annual sales
e. SIC codes
3. Which of the following was not a feature of the BenefitMall
case study?
a. Broker newsletters
b. Satisfaction surveys
c. SIC analysis
d. Welcome kits
e. Reactivation emails
4. In the Isuzu case study, what percentage of dealers logged on to
the site and ordered postcards?
a. 2 percent
b. 4 percent
c. 6 percent
d. 12 percent
e. 50 percent
5. Which of the following is not a benefit of vendor-managed
inventory?
a. Vendors do the selling for their customers.
b. Vendors own the products in their customers’
warehouses.
c. Vendors can call back slow-moving items.
d. Vendors can predict customer product usage.
e. Inventory costs for customers are down.
6. In the Radisson Hotels case, what did the “hurdle” mean?
a. A goal for total sales
b. A minimum that was the same for all agents
c. A level that only top agents could achieve
d. A rate calculated using the agent’s previous sales
7. In the Isuzu Postcard Direct program, what was the best way
for agents to order the postcards?
a. The Web
b. Emails
c. Direct mail
d. Phone
e. Fax
8. What was the advantage to National Semiconductor of the
FedEx arrangement?
a. National cut shipping costs by 8 percent.
b. National eliminated eight warehouses.
c. FedEx took over some of National’s shipping
responsibilities.
d. National cut delivery time to customers to 1 week.
e. All of the above.
9. Who owns “forward-deployed” products?
a. The supplier
b. The customer
c. The taxpayers
d. The marketing department
e. The IT department
10. Why is the discount rate in year 1 of a business-to-business
LTV table seldom equal to 1?
a. Business marketing costs are higher.
b. The SIC codes are often unknown.
c. Business customers usually pay long after shipment.
d. The annual sales to businesses are greater than those to
consumers.
e. Business databases are usually smaller than consumer
databases.
213
C
H A P T E R
10
Customer
Management
O
nce you have a customer database, you can use it to manage your
customers. What does “managing customers” mean? Marketers
have found that they can
• Identify their Gold customers and work hard to keep them.
• Identify their Silver customers and persuade them to buy their
way up to Gold status.
• Identify the customers who are most likely to defect and get
them to change their minds.
• Reactivate lapsed customers.
• Get customers to refer others, who also become
customers.
• Determine the next best product for each customer, and get
customers to buy the product.
• Manage the second sale to one-time buyers.
• Manage leads to get them to become buyers.
• Develop loyalty programs to maintain customers’ interest.
Most of these things were only dreamed of in the 1980s. Some
companies became proficient at customer management in the 1990s.
Today, using the Internet, database marketers are learning to do all
these things.
214
C u s tom e r M a n a g e m e n t
A Little Restraint
A little restraint is required. Just because, theoretically, you can manage
all your customers using your database does not mean that you should.
Managing the top 20 percent is quite easy. Managing the next 20 percent
is a bit harder. Managing the bottom 20 percent may not be worth the
effort. In direct response, 80 percent is about the most important number. Using this number, you can increase your response and your sales
while lowering your cost per sale. Let’s see how valuable 80 percent is
to you.
Gold customers (the top 20 percent of your customers) bring in
80 percent of your revenue. These people are really your company.
Lose them, and you lose it all. Spend a small amount of money on
retaining them, and you keep 80 percent of your total business. How
does this work?
When you hear people complaining that database marketing is too
expensive, you will find that they are talking about 100 percent. Suppose
you develop a retention program consisting of birthday cards, thank-you
letters, appreciation emails, and special customer services and benefits
for certain customers. If you were to do these things for all your customers, the costs would be quite high. The benefits you would receive
might not pay for the cost of the extra services. But why not concentrate on the Gold customers—those who bring in 80 percent of the revenue but represent only 20 percent of the customers? The cost of a
program for Gold customers is only one-fifth the cost of a similar program for all customers, but it safeguards 80 percent of your revenue.
The 80 percent rule also applies to your database cleanup. Your database is a mess. Many of the names and addresses are wrong. Many
entries lack phone numbers or email names. Cleaning it up will be
costly. But wait a minute. Why clean it all up? Why not clean up only
the easiest 80 percent? Your programmers will tell you that those 80
percent can be cleaned up in an afternoon. Getting the remaining 20 percent correct will take several months. Go ahead and clean up the 80
percent right now. Get it done. Then clean up the remaining 20 percent at your leisure, in the most inexpensive way possible or, better yet,
215
216
apply the 80 percent rule to them. What does that mean? Run through the
remaining 20 percent of bad addresses and pick out the Gold customers.
Clean those up. As for the rest, forget about them for now.
You want to send out emails to your customers. But some of them
are on AOL and can read only text. Others can read HTML. Still others
want their copy in Spanish. It is a big job, costing a lot of money and
resources. How can you solve this problem? Use the 80 percent rule.
Pick the easiest 80 percent of your customers and get your message off
to them right away. If you have a consumer database, AOL probably
represents 80 percent. If it is a business-to-business (B2B) database, 80
percent can probably read HTML. Design your email for the 80 percent
and get something done now. If the email is a success, then figure out
how to reach the remaining 20 percent. If it is a failure, don’t bother
with the remainder. Either way, you have kept your costs down.
Building a Data Warehouse
Building a data warehouse for 100 percent of their customers and those
customers’ data has proved to be very expensive for many companies. Most
large companies that have built such a warehouse have spent at least $20
million on the project. Why is it so expensive? Because a data warehouse
is designed by a committee. It has to include all the data that an organization ever conceivably uses: data on customers, prospects, leads, transactions, promotions, responses, products, prices, models, and employees.
Since the accountants are involved, the data in the warehouse have to
balance to the penny. Since IT is involved, the warehouse is built by inhouse programmers who have never built a warehouse before; they are
very excited by the idea, but they have no experience at all.
As most people who have studied data warehouses will tell you, these
warehouses almost never pay for themselves in terms of increased profits.
Why not? Because the benefits of using a warehouse are incremental.
The warehouse will enable you to do better targeting of your communications. Instead of getting a 2 percent response rate, you will get a
2.1 percent response rate. The question is, will the profits from the 5
percent increase in the response rate pay for the cost of the warehouse?
In most cases, they won’t, and you can prove it to yourself with a simple Excel spreadsheet before you begin to build your warehouse.
Suppose your company is taking in $156 million per year in direct
sales to customers. The benefits of a warehouse are shown in Table 101. That’s very nice. Your warehouse has brought in 30,000 more sales,
totaling $7.8 million more per year. But let’s look at the incremental
profits. Suppose you are making a 30 percent profit on direct sales. Table
10-2 shows what would happen.
Your warehouse will lose you about $1.1 million per year. So what
should you do? Use the 80 percent rule. Don’t build a data warehouse
that will retain all the data you could ever possibly need. Build a simple
data mart that meets 80 percent of your marketing needs. Even the
largest corporation in America can build such a data mart for less than
$1.5 million total, with an annual cost of about $800,000. Table 10-3
Table 10-1 Warehouse Costs
Number of offers
Response rate
Number of responses
Sales conversion
Number of sales
Average sale
Sales
Current
Warehouse
100,000,000
2.0%
2,000,000
30%
600,000
$260
$156,000,000
100,000,000
2.1%
2,100,000
30%
630,000
$260
$163,800,000
Increase
100,000
30,000
$163,800,000
Table 10-2 Increased Profits from Complete Data Warehouse
Annual warehouse benefits
Annual warehouse costs
Loss from warehouse
Increased
sales
Profit
rate
Increased
profits
$7,800,000
30%
$2,340,000
($3,500,000)
($1,160,000)
Table 10-3 Increased Profits from Data Mart
Data mart benefits
Data mart costs
Gain from data mart
80%
increased sales
Profit
rate
$6,240,000
30%
Increased
profits
$1,872,000
$ 800,000
$1,072,000
217
218
Table 10-4 Warehouse Costs
Step
Cost
Build warehouse
Annual maintenance
Customer communications
Total
$6,000,000
$2,000,000
$ 500,000
$1,000,000
$3,500,000
shows the result of the 80 percent solution. You have converted a $1 million loss into a $1 million profit increase by using an 80 percent solution.
Where did the $3,500,000 annual cost for a warehouse come from?
Here is a simple way to measure data warehouse costs. If it costs you
$6 million to build the warehouse, you can amortize these costs over 3
years, for $2 million per year. Maintenance of the warehouse will cost
you at least $500,000 per year. Once the warehouse is built, you need
to use it to communicate with your customers—otherwise, what is the
warehouse for? Let’s say that you spend only $1 million on customer
communications per year. Then the warehouse will cost you at least
$3,500,000 per year, as shown in Table 10-4.
Whenever you or any of your colleagues at work come up with a proposal for a large direct response project, Web site, or warehouse, ask
this question: How much could we save by concentrating on the easiest
80 percent and postponing the rest for later? You will be amazed at the
profitability of the answers.
Risk/Revenue Analysis
There is another way of cutting the costs of customer management
besides using the 80 percent rule. That is applying risk/revenue analysis.
For this to work, you have to have done two things:
1. Determined the lifetime value of your customers and divided the
customers into high, medium, and low lifetime value segments.
2. Determined the likelihood that customers will leave you and
divided the customers into high, medium, and low likelihood of
defection segments. When you have done this, you can develop a
simple table that looks like Table 10-5.
Table 10-5 Risk/Revenue Analysis
Lifetime value
High
High
Medium
Low
priority A
priority B
priority C
Likelihood of Defection
Medium
Low
priority B
priority B
priority C
priority C
priority C
priority C
Assuming that all your segments are of equal size, you will have 44
percent of your customers in priority A and priority B and 66 percent
of them in priority C. Customer management is not free. It costs money
for communications, benefits, recognition, and rewards. Why waste a
lot of money on managing your priority C customers? They are in one
of two categories:
1. Of very low lifetime value to you
2. Very loyal and unlikely to defect
In either case, spending a lot of resources trying to modify their
behavior is likely to be a waste of money. This means that you can concentrate your energy on managing (modifying the behavior of ) only 44
percent of your customers, instead of 100 percent. This concentration on
the real customer management problem will pay dividends in many ways.
In the first place, if you try to manage everybody, the resources you
can devote to any one customer will be limited. Let’s say that you have
a customer management budget of $1 million and a customer base of
1 million customers. That is $1 per customer per year. You can’t do
much to change a customer’s behavior with $1 per year. That’s two letters
and an email at best. Suppose you concentrate on only priority A and B
customers. You will have $1 million to spend on 44 percent of your customers, which comes to $2.27 per customer. Even better, if you devote
$4 per customer to priority A customers, you will still have $1.67 left
for each of the priorit y B customers. You can do a lot more with
$1.67 than you could have done with only $1, and much more with $4.
This chapter presents several case studies of different methods of
customer management. All of them look appealing, and they are,
because they have worked well for some companies. How can you
determine whether they would work for you?
219
220
Table 10-6 Increased Return
Leads
Conversion rate
One-time buyers
Conversion rate
Two-time buyers
Annual revenue
Total revenue
Present
With
customer
management
Cost of
customer
management
Net
increased
return
500,000
30%
150,000
50%
75,000
$200
$15,000,000
500,000
36%
180,000
60%
108,000
$240
$25,920,000
$4,000,000
$6,920,000
There is a relatively easy answer. Plug your own numbers into Table
10-6 to determine whether customer management will work for you.
Begin by asking yourself what success you are having today with
• Conversion of leads to buyers
• Conversion of one-time buyers to two-time buyers
• Cross-selling products to existing customers
We are assuming here that you develop a customer management
program that costs you $4 million a year to run. It consists of a database
with communications to the customers and to the sales force that is
managing the program. A successful customer management program
should increase the lead conversion rate, the conversion rate of one-time
buyers to two-time buyers, and, through cross selling, the annual revenue
from two-time buyers.
What will these customer management communications be? This is
up to you. At this point, you know that you want personal communications based on data in your customer database. You want to capture
email names so that most of the communications are essentially free. In
the rest of this chapter, there are several valuable case studies showing
how some companies have been able to manage their customers to
produce results like those shown in the previous tables.
Loyalty Programs
One of the most wonderful ways of managing customers is to set up a
loyalty program. The basic idea is that customers earn some sort of
points or miles based on their purchases, and these points or miles earn
them rewards. The best loyalty programs also include status levels. Customers work hard to achieve various status levels.
Those who have read my previous books know how hard I worked to
become Gold on American Airlines and how happy I was when I
achieved this level. It meant that I got to upgrade to first or business class
whenever I flew. I got to go on the planes first, with the little children
and the old people. I earned bonus miles. I had a special 800 number just
for us Gold folks, and I had Gold printed on all my boarding passes so
that flight attendants and others would treat me with the appropriate
respect. Readers will be happy to learn that because of all these books,
I continue to fly all over the world. I have now become Platinum for Life
on American. For some reason US Airways, which had joint arrangements with American for a couple of years, gave me Silver Preferred.
Now what is the effect on me of these loyalty programs? Well, they
have me hooked. I never fly United or Delta or any other airline if I can
help it. Why would I want to sit jammed into the back of the plane? So
the effect of a successful loyalty program is that you get as close to 100
percent of your customers’ patronage as the customer can get. Now
that’s customer management.
So how do loyalty programs apply to you, since you are not an airline? This is something that’s worth thinking about. Hotels and rental
cars get into the act by awarding airline miles. The effect is not as powerful as it is for airlines, but the loyalty program really helps. Lately
I have always sought out Hilton hotels because they let me double-dip. I
get both air miles and Hilton points.
But most of the readers of this book are not in the travel business and
want to know whether loyalty programs will work for them. I can report
that loyalty programs seem to work in a number of areas, including
•
•
•
•
•
Retail stores, such as shoe stores, dress stores, and department stores.
Supermarkets.
Car washes, auto lubes, and dry cleaners.
Hallmark Gold Crown stores and similar establishments.
Banks— by rewarding customers with high or increased balances
and holders of multiple accounts.
• Business-to-business firms.
221
222
• Restaurants.
• Telephone companies, cell phone providers, and long-distance
carriers.
• Credit card providers. I have three credit cards from Citibank that
award American Airlines miles. I use them instead of cash or
checks whenever I can.
What do you need in order to get started?
• A valid idea for some easy way to award points when customers
spend money with you. This may take some doing in the retail area.
At present, most people’s wallets are crammed with various plastic
cards. Few people will want to add more. How, then, can you award
points? My suggestion is the iDine method discussed in Chapter 9:
by getting people to register their credit cards with you. Then you
can award them points whenever they use these cards with you.
• Some practical plan for providing benefits. There are really two
types of benefits:
• Immediate rewards, sometimes including preferred treatment
• Long-term rewards (like air miles), where the customer is
saving toward an achievable long-term goal
• A lifetime value analysis that permits you to determine in advance
whether the plan will make you more profitable or make you go
broke. Some stores, like supermarkets, operate on very thin margins.
They can’t afford to give away anything substantial. Yet they have
made loyalty programs work very well for them, as we will see.
• Some sophisticated software that maintains the points and
redemptions, handles monthly reports to the customers, and
reports to you. In most cases, companies have outsourced their
loyalty programs. You really should work with a firm that has been
through this process before and knows what can go wrong.
• A Web benefits microsite where customers can come to see what
they have earned and redeem points for rewards. Try, if possible,
to use this Web site for your loyalty program instead of live
operators. The customer service costs can eat up the profits from
the program if you are not careful.
• Reports to the members that are made available on the Web, or
sent by email if your customers prefer it. If you send monthly
statements, your report can be printed on the statement. Airlines
send statements through the mail and have found them to be quite
costly. They are all trying to shift to Web-based reporting.
• A clear expiration date that is announced when your program
begins, so that if it is not working out, you can gracefully exit. You
can always extend the deadline, but you can get in real trouble if
you try to cancel a program while many customers are holding
onto hard-earned points.
• A promotion plan that aims at a very high utilization rate. The
place to start is with your employees. Enroll them in the plan
from the start—with special benefits. You want all your employees
to look the customer in the eye and say, “You really should be in
this program. It is great. I am in it myself.” Promote the plan in
every possible way. The purpose of the plan is to modify customer
behavior. You can do that only if customers are enrolled.
• Status levels, if appropriate, with appropriate status rewards. The
reason for status levels is to get customers to work hard to achieve
them. The Platinum program for American was 50,000 miles in a
calendar year. I used to go crazy trying to meet this goal. I would
take two legs on an American flight when a direct flight was available
on other airlines just so that I did not end the year with 46,000 miles.
Let’s look at some examples of nonairline loyalty programs.
Supermarkets
Most supermarkets today give their customers plastic frequent shopper
cards. These are scanned at the checkout counter and permit the supermarket to build a valid database, giving it information on when and
where you shop and how much you spend in each department. How do
the supermarkets reward loyalty? There are several ways:
• Few of them provide miles or points. The margins in
supermarkets are too thin to allow them to give anything away.
223
224
• Many of them use straddle pricing. They price many items in the
store a small percentage higher than the price charged at other
stores, but they let their members have the items at a small
percentage below the market price if a loyalty card is used. These
items, and there can be hundreds of them in a store, are usually
marked with their occasional shopper price and their member
price. Some of these member-priced items are things that food
manufacturers are trying to promote; others are just items that the
store management has decided need a lift. The effect of straddle
pricing is to reward loyal customers while making occasional
customers who are not members pay for the rewards. It is a
wonderful system. A nonmember won’t care that much that his
milk costs $1.05 while members pay $0.95. If he does care, he can
always sign up for a card and get the price break then and there.
Members, of course, see their membership benefits at the bottom
of the cash register receipts that are printed out for them: “Today
you saved $4.24 by using your VIC card.” The beauty of this
system is that the store did not have to provide this $4.24. It was
provided by occasional shoppers.
• Stores use their loyalty programs to modify customers’ behavior
in other ways. Members get free ice cream on their birthdays. If
they spend $50 a week in the 6 weeks before Thanksgiving, they
get a free turkey. Members who come in on Tuesday night get an
additional saving, since Tuesday is a slow night.
• Members who shop only in the deli and not in the produce
department will get a special certificate giving them $1 to spend in
the produce department. This will get luncheon shoppers to buy
stuff to take home. This $1 is awarded only to the deli-only
shoppers. Other certificates will get produce shoppers to use the deli.
• Lapsed members who spent a lot but who have stopped buying for
6 weeks will get a special offer to encourage them to come back to
the store.
Specialty Stores
Shoe stores, dress stores, department stores, bookstores, and other types
of stores have used loyalty programs with some success. Here, however,
the process is more complicated and often less rewarding than it is for
supermarkets or travel-related businesses. Let’s take the DSW Shoe Warehouse. I bought four pairs of shoes from its store in Fort Lauderdale. The
store gave me a DSW paper card that I could use to get further benefits.
I have never used it since—after all, I am not the shopper in our house.
Helena is. But what has happened is that I now think of DSW when I think
about shoes. It is because of the card, I think. I am sure that this card does
affect customer behavior in a positive way. The rewards to the retailer are:
• Some people will return and use the cards to buy more products.
• Some will remember the store because of the card, and may buy
more products.
• The store gains a mailing list, and hopefully an email list, that has
real value.
• As we know from this book, the store can calculate the value of
each customer who gets a card and determine whether this covers
the cost of issuing the cards.
Surveys by the Carlson Marketing Group have shown consistently
that while only 40 percent of consumers participate in loyalty programs,
those who do participate have higher-than-average incomes, and, once
they are involved, their behavior is affected by the cards.
Let’s see what a loyalty program could do for a retail store chain.
Table 10-7 gives the picture of customers before the program is introduced. It costs $32 to acquire a new customer who makes an average of
1.4 visits per year and spends $50 per visit. The modest lifetime value
of $3.00 in the first year rises to $29.42 in the third year.
Let’s now plan a loyalty program. We are giving on-the-spot benefits
to members in order to modify their behavior. We are seeking to
• Increase the number of visits per year
• Increase the spending per visit
• Increase the overall retention rate
To get these benefits, we are prepared to spend $5 per customer in the
first year, with the amount increasing to $10 per customer for loyalists
who have been with us into the third year. Table 10-8 shows what could
225
226
Table 10-7 LTV before Loyalty Program
Year 1
Year 2
Year 3
Retention rate
Customers
Visits per year
Spending per visit
Revenue
40%
200,000
1.4
$50
$14,000,000
45%
80,000
1.6
$60
$7,680,000
50%
36,000
1.8
$70
$4,536,000
Cost percentage
Costs
Total costs
50%
$ 7,000,000
$ 6,400,000
$13,400,000
49%
$3,763,200
48%
$2,177,280
$3,763,200
$2,177,280
$600,000
1
$600,000
$600,000
$3.00
$3,916,800
1.12
$3,497,143
$4,097,143
$20.49
$2,358,720
1.32
$1,786,909
$5,884,052
$29.42
Profit
Discount rate
NPV of profit
Lifetime value
Table 10-8 LTV with Loyalty Program
Year 1
Year 2
Year 3
Retention rate
Customers
Visits per year
Spending per visit
Revenue
50%
200,000
1.6
$55
$17,600,000
60%
100,000
2
$70
$114,000,000
65%
60,000
2.4
$80
$11,520,000
Cost percentage
Costs
Database costs
Loyalty program
Loyalty program costs
Total costs
50%
$ 8,800,000
$ 6,400,000
$ 500,000
$5.00
$ 1,600,000
$17,300,000
49%
$6,860,000
48%
$5,529,600
$ 250,000
$8.00
$1,600,000
$8,710,000
$ 150,000
$10.00
$1,440,000
$7,119,600
$300,000
1
$300,000
$300,000
$1.50
$5,290,000
1.12
$4,723,214
$5,023,214
$
25.12
$4,400,400
1.32
$3,333,636
$8,356,851
$41.78
Profit
Discount rate
NPV of profit
Lifetime value
happen. The program has cost us in the first year, but the third-year
profits are substantial. Third-year lifetime value has risen to $41.78.
The total profits from the program are given in Table 10-9.
If you look closely at these numbers, you will see that a loyalty program is not a sure thing. If it works, we will get $2.4 million pure profit
out of it. But for it to work, the number of visits per year has to go up, the
Table 10-9 Gain in LTV from Loyalty Program
Old LTV
New LTV
Change
With 200,000 members
Year 1
Year 2
Year 3
$3.00
$1.50
⫺ $1.50
⫺ $300,000
$20.49
$25.12
$4.63
$926,071
$29.42
$41.78
$12.36
$2,472,799
spending per visit has to go up, and the retention rate has to go up. Will
your loyalty program have these results? You had better test the program
on a small scale at one store to see what will happen before you make a
big mistake.
A loyalty program is not a lead-pipe cinch. For more on this subject,
read Brian Woolf’s excellent book Customer Specific Marketing. A key
point that he makes is that without adequate promotion to the customers
and adequate benefits, the program will fail. This type of program is
really directed at the frequent shopper who is not as interested in longterm benefits, such as a trip to Hawaii, as in immediate rewards.
Automatic Trigger Marketing
IMarket, a D&B company, sells data to companies that want to get
business leads and learn more about their existing customers. IMarket
initially delivered data on a CD-ROM (CD) with a meter attachment
for the user’s PC that permitted the user to select records from the CD
for a given price per name. To use the system, companies had to install
the CD and register with iMarket.
The arrival of the Internet gave iMarket a new channel for delivering
product—by the Web rather than by a CD. Zapdata.com, iMarket’s
Internet product, provided a much wider market for the company’s
product. However, Zapdata.com required the company to develop new
marketing strategies to acquire, retain, and cross-sell products through
both the Internet and the desktop.
IMarket’s objectives were twofold:
• To create a database that would provide a complete view of
iMarket’s customers and prospects, both those using the desktop
and those using the Web.
227
228
• Set up a one-to-one automated marketing program that would
enable the company to understand past behavior and the response to
past promotions, and to gather new information through the Web.
These objectives had to take into account the fact that iMarket was a
small company with limited in-house resources. Therefore, iMarket created an in-house customer and lead information database that was
segmented using
• RFM analysis
• Logistic regressions based on business demographics
• Software that permitted the merging of records from different
units of the same companies and the elimination of duplications
from those records
The resulting segmentation system was used to create 10 meaningful RFM segments: Virtuoso, Prodigy, Student, Toddler, Precocious,
Challenged, Infants, Newborn, Lost Opportunities, and Inactives.
These segments are shown in Figure 10-1.
Figure 10-1
iMarket Prospect Segmentation System
Virtuoso
and Prodigy
Our best customers
spent more than $3000
and bought within last
18 months
Student
Made a third purchase within the
last 18 months
Lost Opportunities
A toddler, student, or
prodigy with no
additional purchase 18
months after last purchase
Toddler
Made a second purchase in the first 18 months
Precocious
Purchased
more than
starter kit
initially
Newborn
Just purchased
starter kit
Inactives
No additional
purchase 18
months after
initial purchase
Challenged
Infants
No additional purchase
in 6 months after
starter kit purchase
No additional
purchase in
12 months after
starter kit purchase
Table 10-10 Prospect Report
Total
Active
Sent
Opened
Responses
to
questions
5516
300
430
2195
792
207
170
142
15
209
123
772
5
50
0
368
3
10
0
149
137
14
147
0
0
Conversations
Zap registrant surveys
Zap registrant nonbuyers
Buyers invited to training
CSR survey
Corporate billing
contract usage
Messages
Automated marketing programs were set up to move customers into
higher-value buying segments. The programs consisted of a large number of dialogs, which were directed and monitored by Revenio’s Dialog
software. A sample report is shown in Table 10-10.
The database provided the sales force automation software with lists
of customers and prospects to contact, indicating the segment they were
in as well as demographics and model scores and the prior history of all
calls and contacts. The communications included a variety of different
messages, such as:
•
•
•
•
•
•
•
•
•
•
Welcome letter
Welcome call
Training fax
Customer support survey
Meter credit promotion
Renewal fax
Nonrenewal fax
Upgrade promotions
Inactive revival
No-contact call
IMarket set up automated trigger marketing campaigns for the
contacts in its database. These contacts had “raised their hands” by
attending one of the company’s seminars, registering at the company’s
Web site, responding to an ad, or visiting iMarket’s booth at a trade
show. Once they are in the database, the company markets to both leads
and customers through phone, fax, and email. Customers can choose
their preferred delivery medium.
229
230
The new database program was used to
• Eliminate duplicate records of customers and leads
• Mailing costs were lowered.
• Customers were no longer treated as leads.
• Different salespeople were no longer assigned the same
companies.
• Fights between salespeople over leads were reduced, and
commission payments were lower.
• The entire company had increased confidence in the quality
of the data.
• Have a unified view of customers across online and offline channels
• There was a better understanding of customers’ needs.
• The sales force was now able to view the customer’s entire
history and sell more confidently.
• Cross-selling opportunities increased.
• Unified company reporting became available.
• Create automated marketing programs
• Batch marketing was replaced by messages triggered by
customer buying behavior.
• The “traffic cop” feature prevented customers from
receiving different messages at the same time.
• Instead of being based on the marketing department’s
timetable, programs were based on customers’ buying and
demographic characteristics.
• The number of programs run was no longer based on the
number of marketers available, since the technology could
add new programs easily.
As a result of the new system, iMarket
1. Increased its revenue by 50 percent per quarter and increased the
value per customer by 35 percent per quarter for seven quarters in
a row.
2. Decreased the cost of customer acquisition by 30 percent per
quarter over seven quarters.
3. Reduced the marketing staff while creating 17 new programs that
were executed daily.
Building Loyalty for a Lottery Program
Golden Casket Lottery Corporation Limited is a world leader in providing innovative lottery games. It has offered entertaining lotteries in
Queensland, Australia, for 85 years and today operates the popular games
of Wednesday and Saturday Gold Lotto, Oz Lotto, Powerball, The Pools,
Casket, and an extensive range of Instant Scratch-Its tickets. The games
are sold through an agency network of approximately 1150 retailers.
Golden Casket’s loyalty program is called Winners Circle. The program has more than 1 million unique Winners Circle members, or
38 percent of the adult population of Queensland. Players use the
Winners Circle card to register their Lotto purchases for the benefits
of security (unclaimed prize checks are mailed to them), convenience
(the card can store up to 20 sets of their favorite numbers), and reward
(every $1 spent gives them another chance to win the weekly $10,000
bonus draw). The loyalty program captures 80 percent of all Lotto
revenue and provides Golden Casket with invaluable marketing information. The gaming market in Queensland is very competitive, with
scores of gaming options available—casinos, gaming machines, horse
racing, sports betting, keno, and so on.
Analysis of Golden Casket’s customer database showed that its
approach to branding was eroding player loyalty. The brands were competing with one another for the same customers, which undermined
player loyalty to the corporation as a whole. Golden Casket asked Acxiom
to build computer models that would assist Golden Casket in building
customer loyalty.
Acxiom built three customer models, which are now rescored every
quarter. They are
• Future value model
• Using a neural network, 2 years’ worth of data, and 133
variables, Acxiom built a model that predicted the combined
231
232
future value of 96.2 percent of all customers on the database
with 96.8 percent accuracy.
• Acxiom allocated the customers to low, medium, and high
future value bands. Using these segments, Golden Casket:
• Created a retention strategy for customers whose
future value was likely to decrease.
• Continued the current strategy for those whose
future value was not likely to change.
• Took a stronger focus to build a relationship with
customers whose future value was likely to increase.
• Behavioral segmentation model
• Acxiom clustered Golden Casket’s customers within value
bands.
• Thirteen behavioral clusters emerged. Acxiom found that six
of these variables accurately defined customer behavior:
• The number of games played
• The number of certificates used
• Brand
• The total amount spent
• The number of games played per week
• Methods of play chosen
• This model told Golden Casket which customers could be
influenced to move to a more desirable cluster.
• Cross-sell model
• Using a neural network to track historical behavior, each
customer was scored with a percentage probability of
playing another of the seven games. Table 10-11 shows the
way it worked (all names are fictional).
• Since Jane Smith is 95 percent likely to play Saturday Gold Lotto,
the strategy is to reinforce her behavior. Trying to cross-sell other
Table 10-11 Lotto Results
Likelihood of Playing These Games
Saturday
Wednesday
Tuesday
Thursday
Gold Lotto
Gold Lotto
Oz Lotto
Powerball
Jane Smith
John Brown
Pat Johnson
95.0%
23.0%
65.0%
2.0%
25.0%
31.0%
2.0%
27.0%
3.0%
1.0%
25.0%
1.0%
brands to Jane would waste marketing dollars, create annoyance,
and undermine her loyalty.
• John Brown is a good candidate to purchase all brands. His
behavior is always reinforced to retain his loyalty.
• Pat Johnson prefers Saturday Gold Lotto, but Golden Casket
should try to cross-sell her on Wednesday Gold Lotto at every
opportunity.
How the Program Worked in Practice
The entire loyalty program was restructured and relaunched around the
cross-sell model. When players purchased an entry, each receipt gave
them a chance to win a free surprise game instantly at the point of purchase. Which free game they could win was determined using the crosssell model (e.g., Jane Smith would win a free Saturday Gold Lotto game,
whereas Pat Johnson would win a Wednesday Gold Lotto game).
The terminals at the 1150 retailers were linked to the central Golden
Casket database. When any of the 710,000 active Winners Circle players
purchased an entry, a personal message was printed at the top of the
lottery receipt. If the cross-selling model indicated that the player was
likely to play Powerball, the message would promote Powerball.
The personal message was also used to alert flag players to outstanding prizes. Before this system, Golden Casket had mailed more than
10,000 checks a week, at a cost of $520,000 per year for the mailing
alone. With the new system, special messages printed on the receipts
alerted winners to their prizes, which could be claimed at the retail agent.
A screen message on the agent’s terminal also used the cross-sell
model on each customer, prompting the agent to either cross-sell the
appropriate game or reinforce the customers’ Winners Circle buying
habit. If the player was to be cross-sold Powerball, the agent would be
prompted to mention the Powerball jackpot for that week.
The cross-selling recommendation was put on the call center screens
as well. When a player called, a screen would prompt the call center
operator as to what game to suggest. The model was also used to
develop direct-mail cross-selling offers. One example was a letter and
promotional premarked certificate that was mailed to 50,000 players
233
234
identified by the models. Of those who were sent the offer, 21 percent
played the game identified by the certificate, compared to 1 percent of
the control group (10 percent of the file). Interestingly, the promotional
certificate was available in the store to the control group. Prior to using
the model, Golden Casket averaged response rates of 10 to 12 percent.
The company now consistently achieves direct-mail response rates of
over 30 percent by using combinations of the models.
The Golden Casket Web site was upgraded to allow Winners Circle
players access to their personal information. Players can check for outstanding prizes, update their details, and opt in and out of various offers
such as, “Email me every time the Powerball jackpot reaches $5 million.”
Additionally, when a Winners Circle player visits online, the cross-selling
table is used to populate the banner ad and promotional fields.
Selecting Media Using the Model
Every television program, radio program, newspaper, and magazine in
Australia is profiled by the Roy Morgan research survey. The survey provides Australian companies that buy mainstream media with the Mosaic
geodemographic profile of customers of each media type and program.
Golden Casket used the behavioral models to select target markets by
media zone. These target segments are geodemographically profiled
against the Roy Morgan survey to create a ranking that gives a “best fit”
to “worst fit” portrait to assist Golden Casket in media buying. Using
this system, Golden Casket determined that radio station B105 in the
9:00 a.m to noon hours was the best geodemographic match for a Powerball ad. For every 1 percent increase in media-buying effectiveness,
Golden Casket was able to save about $100,000. This new approach
changed the way Golden Casket bought advertising. The company
determined that in regional areas, less expensive off-peak television could
be purchased without compromising reach goals for specific brands.
Results of the New System
Golden Casket is now more focused on managing customer relationships. Instead of casting a wide net, customer models have enabled
Golden Casket to focus on customers who are more likely to purchase
its products. The Web is used to reduce costs via self-answered queries,
where normally it would cost Golden Casket more than $7 to answer
each call. Web visits are virtually free.
Using the models has tripled Golden Casket’s incremental revenue
from direct mail with cell sizes of 50,000. It is anticipated that going to
smaller cell sizes and using email will continue to increase these
response rates.
Golden Casket’s original rewards loyalty program was complex and
generic, with an annual fee of $2. Only 486,000 of the 723,000 active
players participated in the program. Using Acxiom’s computer models,
the rewards program was restructured and the annual fee was increased
to $3. Additional features, including weekly $10,000 bonus drawings
and surprise games, were added, and check fees were abolished. There
are now 710,000 players participating in the program, and the relaunch
generated $1.1 million in incremental income, which was equivalent to
gross profit, as no cost of goods sold resulted from the change.
Overall, customer modeling and database marketing have allowed
Golden Casket to be more targeted in its communication, to strengthen
player loyalty, to increase the efficiency and effectiveness of media buying, and to reduce costs in the business.
Life Matters
Scotiabank, based in Toronto, Canada, launched an integrated one-to-one
marketing program called Life Matters in a series of “waves” targeted
at specific customer groups. Each wave of this program, which was the
brainchild of Jonathan Huth, vice president, relationship database marketing, included a bundled marketing package with varying financial
product offers. Customers were selected on the basis of their product
holdings, credit criteria, and demographic data in Scotiabank’s data
warehouse plus predictive models. The initial direct-mail message was
followed by a personal call from a branch officer or call center representative. The campaigns were supported through all possible channels,
including direct mail; mass advertising through TV, newspapers, radio,
235
236
and online; point-of-sale materials at branches; outbound telemarketing; and personal contact by branch personnel.
The Canadian consumer marketplace for financial products was
highly competitive. Aggressive marketing by single-product “monolines,” particularly credit cards, had increased the amount of near-junk
mail that consumers were receiving. Capturing and increasing the
scope of business from existing customers (rather than trying to
acquire new customers) was the goal of the new program. The objectives of the new program were to:
• Increase bookings and profits from database marketing campaigns.
• Reduce the cost of campaign creation and mailings, and increase
response rates.
• Automate multisegment and multichannel campaigns, while having
the flexibility to implement rapid changes in design and targeting.
• Enable the database marketing group to manage entire closedloop campaigns, including tests and controls and measurement
of results.
• Make effective use of the bank’s data warehouse.
• Bundle relevant offers in a single but varying package, reducing
clutter for the customer and thus becoming more customer-focused.
The bank began the Life Matters campaign with a test of 50,000
customers to prove the concept. Response rates of 25 percent led to
acceptance of the idea. As a result, several additional bank business lines
asked to be included in the second wave.
The second wave was aimed at an additional 565,000 bank customers, divided into 60 test and control cells. The mailings were done
in two drops to take call center capability into consideration. The third
wave reached over 800,000 customers, divided into 750 cells. The
fourth wave was also aimed at 800,000 customers, but they were divided
into only 270 cells. This last wave included more sophisticated modeling that integrated each individual customer’s propensity to purchase
each of the products or services being offered within the package and
the expected profitability of each new sale to the bank.
To illustrate the technique, in the Life Matters 2 campaign, the offers
were the following:
• Each customer was offered a preapproved line of credit based on
that customer’s creditworthiness. Since these customers were
preapproved, all they had to do was to accept the offer without
further credit adjudication. The offers had credit limits ranging
from $5000 to $15,000.
• Customers were invited to switch their deposit accounts from
another financial institution automatically, with fees waived for
3 months.
• New term deposits were offered with different rate incentives,
based on the profitability of the targeted customer.
• The wave tested different mortgage offers to determine which
would pull the best.
Customers to be included in the program were pulled from the bank’s
data warehouse of over 6 million retail customers. The candidates were
selected on the basis of an estimate of their need for the products included
in the offer. In Life Matters 2, for example, customers who did not own
the products being offered were selected. In Life Matters 4, the prime
criterion for selection was the potential profitability of the account.
How the Campaign Changed the Bank’s Marketing
Approach
Prior to implementing Life Matters, the bank’s database marketing campaigns had primarily been focused on a single product. It was possible
for a customer to be repeatedly contacted with several product-centered
offers that were sometimes confusing and frustrating. The new system
allowed the bank to communicate with customers in a more customerfocused manner, while at the same time reducing the costs associated
with running many single-product campaigns. Bundling the offers
allowed the database marketing group to balance customer focus with
the objectives of various product lines and channel capacity constraints
for lead follow-up.
As a result of the lessons learned through the successive waves, the
bank was able to develop complex campaigns approximately 25 to 40
percent faster than by using the previous SAS-based approach. Annual
marketing program cost savings of approximately Can$3.5 million were
237
238
achieved through eliminating separate product campaign mail and
branch contact costs.
The cost and time savings enabled the database marketing team to
execute campaigns for other divisions, such as Wealth Management,
Insurance, and Small Business.
In total, 120 percent more marketing campaigns were run than had
been run in the previous 12-month period, resulting in almost 9 million
customer contacts—an increase of 180 percent. Life Matters campaigns
yielded response rates of over 20 percent and produced an average
return on investment of over 100 percent.
Profitable Use of Customer Data
The Union Bank of Norway (UBN) had 1 million customers and $25
billion in assets. During the last 20 years, as the bank had grown and
become more automated, it had become clear that bank employees were
not interacting directly with customers as often as they had been, which
meant that the bank was losing touch with its customers’ needs. The
bank was still conducting traditional direct marketing programs with
some success, its highest response rate being about 10 percent.
Since it operated multiple computer platforms with data scattered
among the 175 branches, the bank could not easily gather and use customer-specific information. Getting a full view of a group of customers took days. Bank management wanted to store and access
customer-specific information in a responsive, easy-to-use format. The
goal was to improve revenue and margins through cross-selling and
increasing customer use of its most profitable products.
To solve these problems, the bank turned to NCR’s Teradata Division, installing a Teradata data warehouse in 1995 to capture and store
all information about customers, including transactions going back 4
years and about 40 data elements from customer survey responses. In
1999, UBN added the Teradata CRM solution, a relationship optimization and campaign management application, for its events-based
marketing initiative.
The Teradata system became the foundation for UBN’s loyalty program and event-based marketing initiative. It helped the bank determine
whom to contact, as it prioritized customers, looking at each customer
and evaluating whether that customer needed to be contacted.
Events were detected by a set of business rules that searched the database nightly or weekly to determine if there had been any significant
changes in a customer’s life. For example, if a customer made a large
deposit in a low-interest account, the system alerted a banker to contact
the customer and let the customer know that he or she was losing money.
Before implementing its event-driven marketing approach, the bank
was concerned that customers might be sensitive to the “Big Brother”
effect. Customer feedback from the pilot program showed that the fears
were unjustified and proved the relevancy of both the offer and the
quality of the interactions.
Result of the New System
By using the data warehouse for direct marketing, UBN increased cross
sales, measured against controls. The bank computed a payback period
on the investment of less than 12 months. Compared to a previous 2 to
5 percent response rate, traditional campaigns using information in the
data warehouse produced conversion rates of up to 40 percent. Loyalty
programs gave UBN response rates of up to 70 percent. A pilot program
for event-based marketing yielded conversion rates of up to 60 percent.
The 60 percent sales conversion rate includes the customer history
picture provided by the new database. Some of those who were contacted did not express an immediate interest. Their response was
recorded in the database. Later, the customer ended up purchasing the
additional products from self-generated decisions or as a result of an
automated follow-up. The new conversion rate represented an 1100
percent improvement over the bank’s previous average close rate for
direct marketing campaigns.
Dealing with Churn in Cellular Service
Cellular communications firms have been losing 2.9 percent of their
customers each month, or up to 35 percent per year. In addition,
companies have to disconnect many customers for nonpayment of
239
240
bills. Disconnecting the wrong customer can translate into significant revenue losses, while keeping delinquent customers also affects
profitability.
A wireless service provider was formed by combining the customers
of 11 different companies within a short space of time. Its first job was
to create a unified database covering the more than 20 million customers that had been in 11 different databases. Rather than waiting years
for a data warehouse, the provider created an immediate temporary
marketing data mart that combined the billing data from each company.
This data mart was accessed by SAS products.
The data mart allowed managers in 35 different markets to analyze
the data, giving each manager the ability to handle customer trends on
a local level. Local managers could immediately address churn, instituting local incentives to keep customers or to rework their existing
contract.
Examples of the tactics used to reduce churn were initiatives for:
•
•
•
•
•
•
Rewarding customers by donating to the Special Olympics
Reduced rates for particular customers
A special deal on a new phone
Free movie rentals at Blockbuster Video
Price plans developed after analyzing customer behavior
Phone trade-in programs
Overall, the wireless service provider wanted to know why customers were leaving, broken down by
•
•
•
•
Amount of time with the company
Minute use
Price plan
Thirty other specific market-level and company-level trends
Armed with these objectives, the wireless service provider reorganized its IT department and hired a dozen SAS programmers to construct a decision support system. When the temporary database was
built, it installed several SAS products on its Sun Enterprise servers,
including Enterprise Miner and AppDev Studio.
Results of the First Year
As a result of the analysis, the company reduced its churn rate significantly compared to a number of control groups selected to match each
of the the test groups receiving the promotions. Even a 1 percent reduction in churn rates produced several million dollars in additional revenue. Several valuable lessons were learned:
• Old equipment turned out to be a top reason for churn. As a
result, the company strengthened its trade-in program and allowed
customers to get a new phone every time they signed a new
contract.
• The database enabled the company to determine which delinquent
customers should be kept and which should be dropped based on
modeling and measuring the future profitability of each.
• The company was now able to make educated predictions about
customer behavior, saving the company millions of dollars per
year. Decision making no longer relied on educated guesses as to
what would draw and hold customers, but was based on hard facts
and figures that provided a detailed map for marketing strategies
to follow when creating campaigns.
In the process of creating the database, the wireless service provider
achieved some valuable benchmarks:
• Reports could now be produced in days instead of weeks, and
reports that were previously impossible could now be created.
• The company could now produce both companywide and marketlevel analysis of customer behavior.
• New fields had been added to its database. These new fields gave
the company the power to make better predictive decisions.
• Through customer profiling, the company was able to tailor
messages to different customer groups more effectively.
Getting Customers to Join a Club
YourSport (a fictitious name for a real company) is an enterprise built
around active sports participants and fans. The sports include, among
others:
241
242
•
•
•
•
•
Football and rugby
Tennis and cricket
Motor sports
Skiing, sailing, and swimming
Bicycling and boxing
To publicize and generate enthusiasm for these sports, YourSport
broadcast sports events and maintained a very active Web site with
news about sports on several continents in several languages. The Web
site included news about teams, chat rooms, fan clubs, videos, TV
schedules, and the ability to shop for clothing and equipment used in
the sports or by fans.
YourSport established a club that allowed its members to chat with
their favorite sports stars and get in touch with their favorite YourSport
commentators through message boards. From newsletters to free email,
members could create their own sports Web sites and stay tuned to the
latest on-air action through the YourSport TV schedule. The club,
which had a nominal one-time membership fee, proved to be a major
marketing success.
YourSport had had a fulfillment database for many years, but it had no
customer marketing database. It knew what its sales were, but it knew
little about its customers and their behavior. This could be learned from
a marketing database. To build and maintain its customer marketing
database, Scott Andrews of YourSport turned to Karen Coolbaugh at
CSC Advanced Database Solutions. The information maintained on this
database enabled Scott to do some highly professional analytical work
to determine the value of the YourSport customer. Scott had access to
the database via the Web using the CSC-provided software Brio. Using
this software, he was able to make some profitable marketing decisions.
Scott found several useful ways to segment YourSport customers:
• Sports team members and nonmembers together equal all buyers.
• Club members and nonmembers together equal all buyers.
By segmenting customers in this way, Scott was able to draw some
valuable conclusions:
• Team and club members were very strong contributors who
converted to multibuyer status at high rates regardless of time
on the buyer file.
• In their first year, team buyers contributed at a rate that was five
times that of nonteam buyers. Within the same time frame, club
buyers contributed at a rate that was eight times that of nonclub members.
• In the second year, team buyers again contributed at a rate that
was five times that of their nonteam counterparts. Club members
contributed at a rate that was 11 times better than that of their
nonclub counterparts.
• YourSport recouped most of the $12 maximum investment to
acquire a new buyer, on average, within the first 9 months of a
buyer’s first purchase. On the average, a buyer who was on file for
2 years contributed approximately $39.
• After being on file for 1 year, club members contributed $71, and
team buyers contributed $89.
• Within 2 years of their first purchase, 81 percent of club members
had converted to multibuyer status. In this same time frame, only
41 percent of the average customers converted to multibuyer status.
• Team and club buyers contributed roughly the same amount
during their second year as they did during their first.
The difference between club or team members and nonmembers,
shown in Table 10-12, was striking. In terms of conversion rates from
one-time buyers to multibuyers, Figures 10-2 and 10-3 show how the
team and club members ranked.
Table 10-12 Segment Performance during
the Second Year
Conversion Orders per
percent
customer
Profit per
customer
Team
Nonteam
75.6%
43.4%
2.46
0.68
$169.34
$18.09
Club
Nonclub
80.5%
39.1%
2.32
0.43
$127.63
$7.30
243
244
Figure 10-2
Team Performance in Conversion to Multibuyers
80.00%
70.00%
60.00%
50.00%
40.00%
75.60%
30.00%
43.40%
20.00%
10.00%
0.00%
Team
Nonteam
Resulting Marketing Decisions
The database permitted YourSport to determine the customer lifetime
value of several customer segments using this same approach. With that
information, the company was able to make strategic marketing decisions that were of great value to the enterprise. Included in these
decisions were the following:
• Keep a maximum marginal investment in consumer lists of $12.
Previously there had been no way to determine what that
maximum would be.
• Set the maximum marginal investment in team-specific marketing
efforts at $30.
• Since club members were so valuable, provide incentives to increase
club membership, whereas previously, there had been no incentives.
• Send targeted emails to people who were not club members,
encouraging them to join.
• Add membership rewards, such as travel benefits, to the club to
encourage increased membership.
Figure 10-3
Club Performance in Conversion to Multibuyers
90.00%
80.00%
70.00%
60.00%
50.00%
80.50%
40.00%
30.00%
20.00%
23.40%
10.00%
0.00%
Goal club
Nonclub
Additional Analytical Steps
Any good marketing initiative, such as the customer value study, always
leads not only to marketing recommendations, but also to ideas for further analysis. This study suggested analysis of
•
•
•
•
•
•
Purchasers by products bought in the first order
Purchasers by brand of merchandise
Multichannel buying behavior
The value of incentive purchases
LTV by region of the world and within U.S. markets
Multicompany and multisport purchasers
The Value of the Database
Today YourSport is miles ahead of its position before CSC built its
customer marketing database. It understands its customers better, and
it has been able to use that knowledge to increase sales, cross sales,
and retention. The database has paid for itself many times over.
245
246
What Works
• Using the 80 percent rule to simplify customer management
• Using risk/revenue analysis to concentrate on the key 44 percent
of your customers
• Capturing and using email names at every step in your customer
communications programs
• Determining the possible improvements in conversion rates before
you waste money on any customer management program
• Creating a simple, understandable system for putting customers
into segments (such as those used by iMarket, Golden Casket, and
Scotiabank) and using these segments to guide your customer
management program
• Where possible, using an event-driven approach to customer
management such as that used by Union Bank of Norway
• Running your customer management programs based on the
segment that customers are in, rather than on some overall
marketing plan that does not take customer status into account
What Doesn’t Work
•
•
•
•
•
Building a warehouse for 100 percent of your customers
Trying to manage 100 percent of your customers
Trying to operate without a customer database
Managing a communications program today without emails
Launching a customer management program without first
determining the benefits and costs
• Providing customer management benefits to all customers equally,
rather than to specific segments
Quiz
1. Using the 80 percent rule, you should
a. Concentrate on the top 80 percent of your customers.
b. Clean up 80 percent of your files first.
2.
3.
4.
5.
6.
c. Get rid of the bottom 20 percent of your customers.
d. Build a data warehouse.
e. Do none of the above.
You should spend your marketing resources on those customers
who (best answer)
a. Are most likely to defect.
b. Have the highest LTV.
c. Are priority B.
d. Are not priority C.
e. Have the highest LTV and the lowest likelihood of defecting.
Which of the following was not a feature of the iMarket
customer management program?
a. Welcome letter
b. Renewal fax
c. Birthday cards
d. RFM analysis
e. Customer support survey
The 13 Golden Casket behavioral clusters were defined by all
but which of the following?
a. Total amount spent
b. Methods of play
c. Number of certificates used
d. Brand
e. Geographic region
What was the most important key to Golden Casket’s success?
a. The cross-sell model
b. The Powerball
c. The Winner’s Circle
d. Gold Lotto
e. The Oz Lotto
Which were not among the results that Golden Casket achieved
by using the new modeling system?
a. The annual fee was raised.
b. The company received $1.1 million in incremental revenue.
c. The number of active players increased by 234,000.
d. Many customers were driven to use the Web.
e. Brand managers fought over the more profitable customers.
247
248
7. As a result of the Life Matters program all but which of the
following happened?
a. Thirty-five percent more marketing campaigns were run in
a 9-month period.
b. Customers were ranked by RFM with a maximum score
of 125.
c. Annual marketing program savings were $3.5 million.
d. There were 10 million customer contacts in 9 months, an
increase of 25 percent.
e. There were 48,000 new accounts opened.
8. In the Bank of Norway case study, all but which of the following
resulted?
a. The bank added 2000 new total customers each month.
b. There was a 60 percent conversion rate.
c. The new database produced responses of 70 to 80 percent.
d. There was an 1100 percent increase in the bank’s
conversion rate.
e. The payback period for the database investment was 12
months.
9. The company rewarded its customers with
a. Phone trade-in
b. Donations to the Special Olympics
c. Free movie rentals
d. Price plans based on customer behavior
e. All of the above
10. The company’s benchmarks from the program included all but
which of the following?
a. Reports produced in a few weeks
b. Market-level analysis possible
c. Messages that could be tailored to customer segments
d. New fields added to the database
e. Fields that permitted it to make predictive decisions
C
H A P T E R
11
Letting Them Come
behind the Counter
Tupperware used to be sold through neighborhood parties. They were
fun. You saw everyone in the neighborhood, and you got the opportunity
to buy some kitchen storage containers that you couldn’t buy in stores.
There is an old story about a Tupperware salesman who was all
thumbs. He arrived at a party with a big box of Tupperware products
and a catalog, but he did not seem to know his business. He took each
item out of the box and could not remember its name or its price, or
even what it was used for. He was an embarrassment to the group.
After watching him flounder around for about 10 minutes, members
of the party began to help him. They took items out of his box, looked
them up in his catalog, and figured out the prices for him. Soon, everyone in the room was assisting this helpless man. By the time the party
was over, the man had sold more Tupperware than most salesmen sold
in a week. In fact, this was his standard method of operation. He always
appeared befuddled. He sold Tupperware by getting customers involved.
T
hat is what is happening today on the Web. Instead of the standard “We demonstrate, you watch,” companies are learning to
let customers come behind the counter and figure things out for
themselves. In many cases, this is proving highly profitable.
There has always been a distance between suppliers and customers.
This distance is typified by the counter. The clerk stands on one side,
ready to answer questions and ring up sales. You stand on the other
249
250
side waiting for service. You are not supposed to invade the clerk’s space.
All of this is changing with the Web.
The Web permits customers to get instant access to lots of information that previously was available only through telephone calls or
personal visits. Federal Express was one of the first to start this with its
package tracking system. Today, anyone can enter a FedEx package
number on the FedEx Web site and instantly find out what has happened
to the package at each point in its journey. Before FedEx had this Web
site, you could call a FedEx operator and get the same information—in
fact, you can still do that today. But by going to the Web, you can get
the information faster, you never get put on hold, and you can print out
all the information on your PC printer. It is a wonderful system for the
customer. How is it for FedEx? It is a huge cost saving. Toll-free calls
to operators cost companies between $2 and $7 per call, including the
cost of the call and the cost of the operator. Web visits cost absolutely
nothing once the system is set up and running.
Super Customer Service
You hear people saying that no one is making money on the Web, but
that’s not quite true. Few companies are profiting from sales to consumers over the Web, but hundreds of companies are saving money and
building customer relationships through customer service on the Web.
These functions generate more profits than all direct Web consumer
sales put together.
Marketing and customer service are usually separate departments
within a company. They have separate goals and methods. However,
that situation is changing because of the realization that good customer
service leads to higher levels of retention and sales. There are several
factors driving this change:
• Industry has discovered the value of loyal customers. They buy
more, buy more often, are cheaper to serve, have higher retention
rates, and are more profitable than newly acquired customers.
• Marketers have discovered that there are two methods that it can
use to influence the level of customer loyalty: recruiting the right
Letting Them Come Behind the Counter
kind of customers to begin with, and treating them very well once
they are acquired.
• Excellent customer care is the most important method for
improving customer loyalty. Customer service personnel are the
front-line troops in the battle to win customers’ loyalty.
• To provide good customer care, customer contact personnel have
to be empowered with information and the authority to make
decisions and to act on the customer’s behalf.
In other words, marketing is no longer a matter of thinking up clever
advertising and direct-mail pieces. Good marketing involves everyone
in the company who has contact with customers. Marketing and customer service are coming together.
Customers Are Not All Alike
Some customer service advocates say, “We will give super service to all
of our customers.” This is a lofty sentiment, but a bad idea. Companies
are learning that we must not treat all customers alike, because customers
are not all alike. Some are loyal and put all their trust in us. Others are
indifferent and spend most of their money elsewhere. Many customers
not only are not profitable but cost us money and will never be profitable. Figure 11-1 repeats the graph of the customers of a southern
bank that was shown in Chapter 1.
Why should we spend money trying to retain customers who are
robbing the company of value and hurting the enterprise? What we
really want to do is provide super service to the top two segments so
that we can retain them. They are the lifeblood of the enterprise. We
want to give them services that we could not possibly afford to provide to everyone. If we limited our services to those things that we
could afford to give everyone, the services would be so meager that
they would be meaningless to the recipients and useless for influencing loyalty.
So what are the rules for successful customer service programs?
• Provide good information. Customer service reps (CSRs) must have
access to the customer marketing database. They must know what
251
252
Figure 11-1
Profitability
79.67%
80.00%
60.00%
24.82%
40.00%
15.83%
1.52%
20.00%
0.00%
–20.00%
–21.83%
–40.00%
5%
11%
28%
28%
28%
each customer has bought in the past and each customer’s
profitability. They must have a record of past complaints and
compliments. They must know who the key people are in each
firm and what those people’s opinions about the CSR’s company
and its products are.
• Empower customer service. If customer service personnel are just
message takers, people will not unburden themselves to them.
Customer service personnel have to represent your firm. They
must be able to do what the owner of the firm would do: make
decisions that are in the interest of both the customer and the firm.
They must be delegated the authority to act to solve problems.
• Set up test and control groups. Good customer service is not just
nice to have. It can be highly profitable. However, you must prove
to management that the money spent on customer service is
creating customer loyalty and profits for the firm. The only way
that can be done is to set up control groups so that you can
measure the performance of the test groups correctly. Many
executives will oppose the creation of control groups, saying, “We
must treat all customers alike.” Fight this old-fashioned way of
thinking with all your might. Without control groups, you can
never prove that what you are doing is having the desired effects
and justifies your enlarged customer service budget.
Along Comes the Web
Customer service, however, is expensive. With thousands of calls per
day at a cost of $2 to $7 per call, companies have seen a significant
drain on their bottom line. Customer lifetime value analysis must be
used to determine whether the helpfulness and relationship building is
worth the cost. In some cases, companies have found that it was not
paying its way. Most banks that have done profitability analyses have
discovered that about half of their customers are unprofitable. A good
part of the reason for the lack of profits can be traced to the expense
of customer service. When Federal Express started the system that
enabled customer reps to field questions on the exact status of every
package shipped at any time, it was wonderful. But it came at a significant cost. If every customer called to learn the status of his or her shipments, all shipments would cost at least $2 more than they do. Using
the Web has saved this money.
The Internet Saves the Day
The Internet is gradually replacing the customer service rep. Instead
of paying for the cost of a toll-free call and the salary of a customer
service rep who reads information off a computer screen, companies
are learning that their customers can look at that same screen directly
on the Internet and get the answers themselves. What’s more, customers like it better. What is the saving produced by fielding a call on
the Web instead of on the phone? Between $2 and $7 per call. Banks
are discovering the saving from the Internet. Transactions on the
phone with a live operator cost banks $2 or more. Those on the Web
cost about a penny each.
The big savings, however, come from inviting customers to come
into your company to read the same screens that your employees read.
Customers become part of your company family. This is real relationship marketing.
Every business-to-business company in America (or in the world, for
that matter) can profit by opening up its internal company data and
253
254
making them available to customers on the Web. There are three reasons
for this:
1. The Web site will probably pay for itself in savings on toll-free
calls and operator time.
2. It may increase sales to new prospects (although that is not a
sure thing).
3. It provides a way to recognize and build relationships with
customers that will improve loyalty and retention.
Who is doing this? Customers use Dell Computer’s Web site to
check their order status 50,000 times per week. The same calls to Dell’s
customer service reps would cost Dell $200,000 per week. Dell also
saves several million dollars per year by encouraging 200,000 customers
per week to make their troubleshooting inquiries through the Web.
Customer Service on the Web
Amazon showed us the way: “Welcome back, Arthur. If you’re not Arthur,
click here.” This is wonderful one-to-one marketing. Many companies
have copied this method. They are finding that it works. It is what the
old corner grocer used to do. He stood at the entrance to his store and
said, “Hello, Mrs. Williams. How’s your daughter taking to Radcliffe?”
Most large companies today are spending millions of dollars on customer service. Hundreds of CSRs are answering customers’ questions
and selling products. Many of the CSRs’ functions can be performed
by a good Web site. Look at the work of a CSR. She takes scores of
incoming calls per day. She has a computer linked to the company’s
main server. As customers ask her questions, she manipulates her mouse
and her keyboard to get the answers on her screen. Then she reads the
answers to the customers over the phone. She gives information, and
she takes orders. A typical call to a CSR costs $6.50. In most cases, a
really good Web site can replace a high percentage of the CSRs’ functions. The method is to give the customers the same access to the company server that the CSR has. Figure 11-2a shows what is happening
now, and Figure 11-2b shows the new system.
Figure 11-2
Elimination of the CSR. (a) Current system. (b) New system.
Company server
Customer
CSR
(a)
Company
Web site
Company server
Customer
(b)
With the new system, we have eliminated the CSR and the telephone
call. Instead, customers using the Web site see the same information
that the CSRs see. They place the same orders, using the same credit
cards. One company was spending almost $50 million per year on CSR
functions. It was fielding 6 percent of its queries on the Web. A consultant estimated that if the company could raise the percentage of its
CSR work that was going to the Web site to 30 percent over a 3-year
period, assuming a 15 percent annual growth rate, it could save more
than $20 million per year in 3 years. That is the promise of the Web.
These are numbers that go right to the bottom line.
Profits like this can come about only if customers will use the Web
instead of calling a CSR. How can you get them to do that?
• The Web site has to be as good as or better than the services
provided by a CSR. Study what your CSRs are doing and saying.
Design your Web site to do the same thing, but then go one step
further. Provide customers with lookup functions that are more
sophisticated than anything that a CSR could do.
• The Web site has to be publicized so that your customers know
about it.
255
256
• The Web site has to be personalized. It has to have the same
friendly personal relationship that a good CSR has. Amazon
doesn’t say, “Welcome back, Arthur” just because it is a neat thing
to say. Amazon says it because it keeps Arthur coming back.
• The Web site has to have a live agent function. Studies show that
74 percent of Web shopping carts are abandoned at checkout. The
reason is that at the last minute, people have questions that the
Web site doesn’t answer. You have to put a button on your Web
site that provides a text chat with a live agent. CSRs can handle
four text chats at once. This software is provided by several
companies, including www.liveperson.com.
Once your Web site is as good as a CSR, you can measure your
success by the percentage of your customer service queries made and
orders taken on the Web rather than by a live agent.
The Web is not a panacea. It will support but never replace direct
mail, catalogs, retail stores, malls, brand names, TV, newspapers, magazines, radio, and books. It can be a very important channel, but only if
the company already has other profitable channels for sales that the Web
will supplement. You can profit from the Web, but you have to look in
the right place.
Premier Pages
Dell began putting its marketing and technical information on the Web
in 1994. By 1996, customers could come to the Dell Web site, configure
the computer they wanted, and place an order. However, the orders came
to Dell as emails. Dell had to retype them into its ordering system!
In 1998 the company introduced Premier Pages, which let business
customers tailor what their employees could see and order. Today, the
Dell system is all electronic. Seventy percent of Dell’s sales are business
to business.
In many cases, Dell negotiated volume prices for its best customers.
The way it works is this: A deal is worked out between Dell and the purchasing officer at the target company. Dell agrees to provide the Web
page and a volume-pricing deal. The purchasing officer agrees to give
the special address of this Dell Web page to all employees who want to
buy computer products. It takes Dell only 30 minutes to create a new
Premier Page for a new business customer.
When an employee comes to the site, he registers as being an employee
and purchases whatever he needs, providing the Dell site with the purchase order number. Every month the purchasing officer gets a report
from Dell showing which of her employees have made purchases and what
they have bought. She has complete information about what her firm is
buying from Dell, who is buying it, and what each employee bought.
These reports are very useful for a purchasing department. They also show
the savings that the company has made through its volume deal with Dell.
Dell has set up tens of thousands of these business-to-business pages.
They give Dell a tremendous advantage in business sales. Today, all of
Dell’s competitors are copying the system and are trying to catch up.
The Dell system gives the business customer a lot of information.
Dell has put 54,000 pages of service information on the Web, and this
can be accessed by any customer at any time. The Premier Page lets the
customer know where any given system is in the production process.
Dell prepares for the order’s arrival by sending shipment information
automatically by email. Once the system arrives, the customer can troubleshoot and upgrade it online.
Getting companies to use the Dell Premier Page system is not always
easy. When Dell tried to get Eastman Chemical to use a Dell Premier
Page for ordering, Eastman at first refused, saying it was too much
work. The company had its own internal ordering system and preferred
to send Dell a fax of the order. In response, Dell revised the site for
Eastman to make it conform to Eastman’s system. Eastman now uses
the Dell site for its computer orders.
There are many advantages to Dell from its Web ordering system.
Dell’s statistics show that without a Premier Page it takes five phone
calls from prospects before the company gets an actual order. The first
phone calls are for information and prices. But if the prospects use the
Web site first, Dell finds that it can sell products on the first phone call.
As a result, Dell’s profits on Web sales are 30 percent higher than its
profits on other sales.
257
258
Once the site is established, Dell uses the tracking information to
increase sales. At one point, an IS director checked networking equipment
on his Dell site. He did not buy. The next week, a Dell salesperson
called this IS director (based on the Web statistics) to see if she could
help the IS director with his networking purchases. The basis of satisfactory customer relationships is knowledge, and the use of that knowledge for helpful communications. Without the Web visit, how would
the salesperson have known that she should call on the IS director to
discuss networking equipment? This is effective use of the Web.
Letting the Supplier Come behind the Counter
We have been talking as if the Web facilitates marketing communications between customer and supplier, which it does. But there is another
vital relationship that is also made possible by the Web. You can invite
your suppliers to come behind the counter as well.
In Strategic Database Marketing, I described the super service that
the Marriott Hotels Web site provides to professional meeting planners.
A meeting planner who logs on to this site can find all the Marriott
hotels in any given area, indexed by the services offered. When the
planner picks out a hotel, she can see how many conference rooms are
available and their capacity, and she can see a picture of the hotel and a
map showing its location. Most of this information cannot easily be
described on the telephone. When you see it on your PC in living color,
you can print it out and carry it around. That is super customer service.
In Chapter 12 I report that Lands’ End discovered that customers who
filled out a personal description of their bodies were 70 percent more
likely to make a purchase than other Web visitors. Getting customers
involved, as the Tupperware salesman did, is the route to customer
loyalty and sales.
So, what should you do? What is your role in the new economy?
Everyone who is reading this book has either customers or customers
of clients who could be helped by letting them come behind the counter.
What are the advantages?
• Customers become more loyal if they look up what they need
themselves.
• The costs to the supplier are lower than the costs of the traditional
ordering system. You eliminate the customer service reps, the
phone calls, and the manual typing of orders.
• Errors in ordering are reduced, since a well-designed Web site
provides a picture of the product being ordered and the correct
part number.
How Difficult Is It to Set Up Such a System?
Let’s start with what we know. Every company in America has put its
inventories and its technical information on computers. This has happened in the last 20 years. These data are available to company
employees using PCs or mainframe terminals. In most midsized or
larger companies today, authorized employees can look up almost anything and see it on a computer screen.
The next step may cost some money, but it is definitely achievable.
Get a programmer to make the data on the screens available to customers through the Web. Set up a microsite that major customers can
access using a PIN and a password. Make the site user-friendly, following the three-click rule: Everything that anyone would want to see
on the site should be available in three clicks of a mouse. To do this,
you have to think like a customer: What would I, as a customer, want
to see on this site? Then design the site with that in mind.
Work with your customers to design the site. Make getting information and ordering very easy. Set up their accounts so that they can
get information free and parts or services, which they have to pay for,
with one click (see Figure 11-3 for an example). Amazon is a master at
this technique. Amazon lets its customers look up all their past purchases
through the Web. I have bought so many books from Amazon that I
sometimes cannot remember what I have bought. In three clicks, I can
find out all the books and videos that I have ever bought from Amazon,
including the prices I paid and the dates I bought them.
259
260
Figure 11-3
One-Click Ordering
I looked up my most recent book, Strategic Database Marketing, 2nd
edition, on the Amazon.com site. In addition to the price and a picture
of the cover, there was a summary of the book, 43 sample pages that
visitors could read and print, nine reviews of the book, and a Sales Rank
Number: 5744. Amazon ranks the sales of all books daily, from 1 to
1,000,000 or more. In addition, Amazon reported that my book was
selling very well to employees of NCR, the makers of Teradata computers. What bookstore could possibly provide such interesting and
valuable information?
Set up reports so that you can know on a daily, weekly, and monthly
basis which customers are visiting your sites, what they do there, and
what they buy there. Design the reports so that your customers can see
them as well, print them on their PC printers, or download them into
their own reporting systems.
Once you have a system set up, how do you get customers to use it?
After all, most of them are happy calling your operators on a toll-free
line and wasting your money. Why would they change? Why would they
go to the Web to get information or to order from you? People do what
is in their own best interest, not what is in your best interest. You have
to give them some kind of incentive. What are the ways of doing this?
• For business-to-business operations, work with the purchasing
officer at each customer site. Offer these purchasing officers a deal
that they cannot refuse: If they will persuade their employees to
use the site, you will give the company a volume price break.
• Provide monthly reports to the purchasing officer telling her what
her employees are buying.
• Put the purchasing officers on an advisory panel. Invite them to
come to Hawaii with you once a year to discuss business. You
figure it out. There is some good way to make it worth their while.
• If your site is for consumers, you have to come up with relevant
incentives. The way to start is to figure what you get out of the
system and share some of the proceeds with your customers.
When the site is ready, you will need to send an email to all your
customers like the ones that Boeing uses:
To provide more value and better service to our customers, we have
developed a way to use the sophisticated technology of the World Wide
Web to give customers direct access to our extensive part information
and ordering system. You can now order parts and view part information through a new World Wide Web site called the Boeing Part
Analysis & Requirement Tracking (PART) Page
How do you justify such a system to your management? The software costs are not trivial. Let’s try some numbers. What would it cost
to let your customers come behind the counter? Table 11-1 gives some
sample one-time costs.
This assumes that your products, services, and technical data are
already in electronic form. If they are, $400,000 should enable you to
set up a Web microsite to which customers with an ID and password
could come to see your products and place orders. The reporting software would provide Web reports to the customers. Promotion is the
money needed to let customers know about the system so that they will
Table 11-1 One-Time Costs
Web access software
Reporting software
Promotion
Total
$400,000
$ 20,000
$ 30,000
$450,000
261
262
use it. If we amortize the cost of the new system over 3 years and add
in the operating costs, the results look something like Table 11-2.
Let’s suppose that you have a business-to-business company with
several thousand customers but that you apply the costs of the new system only to your best 1000 customers. We will charge the total cost of
the system against these best customers, even though many others will
probably use it. Table 11-3 gives the base lifetime value of these 1000
customers before we introduce the new system.
On average, each of these 1000 best customers spends $28,000 per
year with you. It costs $400 to acquire each of these customers, and you
are spending $50 per year on marketing costs to maintain relationships
with them. After 3 years, the LTV is over $16,000.
If the new system is successful, it will create two changes in customer
behavior. Those who adopt the new system will be more loyal. The
retention rate will rise from 70 percent to 75 percent. In addition,
Table 11-2 Yearly Costs
Costs
Month
Year
Site maintenance
Amortization of one-time cost
System management
Total
$ 4,000
$12,500
$ 6,000
$22,500
$ 48,000
$150,000
$ 72,000
$270,000
Table 11-3 Base LTV
Year 1
Year 2
Year 3
Retention rate
Customers
Annual spending
Revenue
70%
1,000
$28,000
$28,000,000
75%
700
$29,000
$20,300,000
80%
525
$30,000
$15,750,000
Costs
Operating costs
Total costs
70%
$19,600,000
$
400,000
$
50,000
$20,050,000
68%
$13,804,000
67%
$10,552,500
$
35,000
$13,839,000
$
26,250
$10,578,750
$7,950,000
1.10
$7,227,273
$7,227,273
$7,227.27
$6,461,000
1.19
$5,429,412
$12,656,684
$12,656.68
$5,171,250
1.28
$4,040,039
$16,696,724
$16,696.72
Profit
Discount rate
NPV of profit
Lifetime value
the average spending rate will increase from $28,000 to $29,000.
These are not unreasonable assumptions. The new lifetime value is
Assuming that you replace the lost customers as they disappear, the
3-year payoff rate for the new system for your 1000 best customers
will look like Table 11-5. The 3-year result would be a net profit of
almost a million dollars, after deducting the costs of the upgraded
Web site.
Boeing Spare Parts
Idle airplanes cost airlines thousands of dollars per hour. How many
times have you sat in an airplane waiting for engineers to replace some
Table 11-4 Improved LTV with New System
Year 1
Year 2
Year 3
Retention rate
Customers
Annual spending
Revenue
75%
1,000
$29,000
$29,000,000
80%
750
$30,000
$22,500,000
85%
600
$31,000
$18,600,000
Costs
Operating costs
Web site costs
Total costs
70%
$20,300,000
$ 400,000
$
50,000
$ 270,000
$21,020,000
68%
$15,300,000
67%
$12,462,000
$
37,500
$ 270,000
$15,607,500
$
30,000
$ 270,000
$12,762,000
$7,980,000
1.10
$7,254,545
$7,254,545
$7,254.55
$6,892,500
1.19
$5,792,017
$13,046,562
$13,046.56
$5,838,000
1.28
$4,560,938
$17,607,500
$17,607.50
Profit
Discount rate
NPV of profit
Lifetime value
Table 11-5 Gain from Changes
New LTV
Old LTV
Difference
For 1000 customers
Year 1
Year 2
Year 3
$7,254.55
$7,227.27
$ 27.27
$27,273
$13,046.56
$12,656.68
$ 389.88
$389,878
$17,607.50
$16,696.72
$ 910.78
$910,776
263
264
malfunctioning part? Did you ever wonder where all the parts used at
the hundreds of airfields that our planes fly out of in the United States
come from? One missing $50 part can cost an airline thousands of
dollars. But airlines cannot afford to keep every possible part at every
possible airport from which their planes fly. The answer is to have quick
access to spare parts from the manufacturer.
For many years, Boeing had a customer service parts-ordering system
with operators standing by 24 hours a day. Each operator had a computer
terminal that helped him to locate the needed parts for his frantic customers. It was not easy. There are 2000 suppliers of parts for Boeing
airplanes. There is competition. There are Asian suppliers who make
replacements for most of the parts in Boeing planes. Spare parts for
Boeing airplanes are a $7 billion per year business, and Boeing wants
to get as much of that business as it can. So, several years ago, Boeing decided to let its customers come behind the counter. It gave its
customers the same access to spare parts that its employees have. Boeing’s Airline Logistics Support organization is now an integral part of
Boeing’s customer support operation.
Airline customers can log on to the Boeing customer support Web
site with a PIN and have access to 400,000 parts in Boeing warehouses in Seattle, Atlanta, Los Angeles, Singapore, London, Dubai, and
Amsterdam. The airline customer can find where her needed part is
located, place her order without human intervention, and, if necessary,
send a driver over to pick up the part.
The Boeing Spares organization was the first to offer spares ordering
on the World Wide Web. Now more than 600 customers access the
Boeing PART Page (see Figure 11-4). Boeing provides its customers with
routine, next-day shipment of spares. The Spares organization also offers
the Global Airline Inventory Network, an exclusive service in which
Boeing manages an airline’s supply chain for expendable airframe spare
parts. The network is designed to eliminate costly inventory inefficiencies.
Industry studies indicate that for every $1 of spare parts inventory,
airlines spend an additional 35 cents on inventory holding and materials
management costs. Designed to attack these inefficiencies and costs, the
new Boeing system provides both airlines and their suppliers with a
more efficient supply chain.
Figure 11-4
The Boeing Global Airline Inventory Network
Here’s how the Boeing system works:
• Boeing takes responsibility for purchasing, inventory management,
and logistics for their airline customers’ expendable airframe parts.
• All airframe parts are forward-deployed by Boeing at or near the
airline’s point of use. That means that the parts are stored either in
a Boeing partner’s warehouse or actually at one of the airline
customer’s parts departments.
• Boeing and other suppliers own these airframe parts until they are
actually used. An airline pays for parts only as it uses them. This
significantly reduces the airline’s inventory holding costs.
• A supply chain management system serves as the “command
center” for the network, monitoring the parts use of customer
airlines and allowing suppliers to better forecast demand and plan
production accordingly.
• The existing Boeing spare parts distribution system, with seven
regional distribution centers around the world, is the backbone of
the network. In addition to Boeing’s proprietary parts, these
distribution centers stock many of the other 2000 suppliers’ parts,
creating a single parts-distribution channel.
This system helps airlines reduce costs and frees up significant
amounts of capital that were formerly tied up in inventory. It enables
265
266
suppliers to benefit from the efficiencies that come from better demand
forecasting and using a single distribution channel.
A Word of Caution
The systems described in this chapter may not work for your situation.
The payoffs from letting customers behind the counter can come only if
• The customers actually want to come behind the counter and will
use your new Web site to do so.
• The customers gain something that is of value to them from using
your new system.
• The costs of setting up the system are not too high in relation to
the payback. Some companies have spent millions of dollars on
such systems with no hope of payback. Boeing’s system was very,
very expensive. But the company was building market share in a
$7 billion market.
• The system will have a measurable impact on your customers’
retention and spending rates. How can you know that? You must
run little tests to find out.
• You have tested everything on a small scale before you make big
mistakes.
What Works
• For every piece of information that you have stored in your
company computers, figure out if it would be of interest to some
of your customers. If so, find a way to let them access this
information directly on the Web without human intervention.
Let them come behind the counter.
• Set up special Web sites for your best customers. Provide reports
to your customers on what they are buying and who is buying it.
• Provide volume-pricing arrangements to induce your customers
to use your new Web site for ordering.
• Before you do anything, use the tables in this chapter to determine
whether your new system will be profitable. If it will be, use these
tables to sell your concept to your management.
What Doesn’t Work
• Setting up a Web site for customer access without talking to your
customers first to determine whether they would use it.
• Setting up such a system without first working out the LTV charts
that show what profit you could make from the system. Many, if
not most, such systems will not be profitable.
• Spending too much on a software system without first testing the
concepts on a small scale.
Quiz
1. What was the point of the Tupperware story?
a. Befuddled salesmen are the best.
b. Customers like to participate in the selling process.
c. Tupperware is no longer successful with parties.
d. You should run parties to sell products.
2. Why provide customer service over the Web?
a. It saves millions of dollars.
b. Customers never get put on hold.
c. You can provide data and pictures that are unavailable in
any other way.
d. You can build loyalty.
e. All of the above.
3. Which of the following is not one of the rules for successful
customer service?
a. Provide good information.
b. Empower the CSR agents.
c. Set up test and control groups.
267
268
4.
5.
6.
7.
8.
d. Eliminate toll-free calls so that people must use the Web.
e. None of the above; they are all rules for successful
customer service.
Which of the following is not one of the rules for successful
Web customer service?
a. The Web site must be as good as or better than a CSR.
b. The Web site must be publicized.
c. The Web site must be personalized.
d. You must have a live agent function.
e. None of the above; they are all rules for successful Web
customer service.
Which of the following is not a feature of the Dell Premier
Pages system?
a. Vendor-managed inventory
b. Monthly reports
c. IDs and passwords
d. Volume-pricing discounts
e. 30-minute setup time
What does “forward deployed” mean in the Boeing parts case?
a. The parts are owned by Boeing customers.
b. The parts are warehoused in customers’ warehouses.
c. The text does not make this clear.
d. Boeing does not know where the parts are.
Which of the following is not one of the advantages of letting
customers come behind the counter?
a. Customers become more loyal.
b. Costs are lower.
c. Customers like it better.
d. Errors are reduced.
e. More CSRs are needed.
Prepare a 3-year lifetime value table for a business-to-consumer
supplier who has a retention rate of 45 percent, 55 percent, 65
percent; a cost ratio of 65 percent, 62 percent, 60 percent; an
average sale of $60, $65, $70 per visit; visits per year of 2.1, 2.5,
2.8; an interest rate of 8 percent; a risk factor of 1.5; an acquisition
cost of $46; and an original number of customers of 200,000.
9. Prepare a revised table showing a 4 percent increase in the
retention rate, an increase in the number of visits per year of 0.4,
and a $4 increase in the average spending per visit. Spend $8 per
customer per year on marketing to achieve these results.
10. Using the previous two tables, prepare a table showing the gains or
losses from the new marketing program in Question 9, assuming
that the company continues to have 200,000 customers.
269
C
H A P T E R
12
Can Database
Marketing Work
for Catalogers?
C
atalogs are the fastest-growing and most profitable direct marketing method in use today. The old Sears Roebuck and Montgomery
Ward big books are gone forever. But what has replaced them are 16
billion small specialized catalogs that are mailed each year to millions
of U.S. households. Why are catalogers succeeding today?
• Lightning-fast fulfillment. Remember “allow 6 to 8 weeks for
delivery”? Who can stay in business today saying that? Many
catalogers today even offer next-day service. That means that they
have streamlined their warehouses and have UPS and FedEx
trucks constantly backed up to their doors.
• Massive exchange of names. North America is about the only place
in the world where you can rent the names of mail-order buyers.
In almost every other country in the world, marketers hold
customer names in an Al Gore “lock box.” Helena, my wife, is an
avid catalog shopper. She gets from six to ten catalogs a day, all
year long. She loves them—she reads every page of every catalog
as if they were novels. As soon as she places an order, the cataloger
immediately rents her name to dozens of other catalogers, who
rush to fill our mailbox. Without this universal name exchange,
the U.S. catalog industry as we know it would soon die.
• Universally available credit cards. Not only do all catalogers depend
on them, but all consumers have them. Just about every American
family receives at least two credit card solicitations per week.
270
C a n D ata b a s e M a r k e t i n g W o r k f o r C ata l o g e r s ?
What does this mean for the catalogers? Financial success. The
problem of getting paid for products has almost disappeared.
• Sophisticated telesales. Today’s call center software is one of the
most complex aspects of cataloging. Millions of agents staff
centers on a 24/7 basis. Using Caller ID, the call center software
recognizes previous customers and calls up their purchase history
on the screen before the call is answered. The computer terminal
screens enable the agent to find any product immediately, so that
she can immediately begin to discuss colors, sizes, and prices
effortlessly. If the caller is a previous customer, the agent doesn’t
have to get the name and address a second time.
• No-hassle return policy. Modern catalogers make it very easy to
return merchandise—and the best ones pay the return postage.
Customers may exchange or get a credit or a full refund with no
questions asked! Such service was unheard of a decade ago, but it
has become standard; 20 percent of all items ordered are returned.
Use of Database Marketing
With all these new developments, you would think that catalogers would
be in the forefront of database marketing today. You would be wrong.
While every cataloger creates a house file of customers so that it can
mail them catalogs and exchange their names with other catalogers, few
catalogers have a marketing database system. Recent studies show that
catalogers with sales of more than $100 million per year spend an average of $18,000 on database marketing. That does not pay the salary of
even a half a person to do database marketing. Those with sales of less
than $100 million spend almost nothing in this field. Why not?
To explain this, let’s define what we mean by database marketing. In
modern DBM, a company keeps track of customer purchases and crafts
special customer communications that are designed to build retention,
referrals, loyalty, and cross sales. Why aren’t catalogers doing this?
The explanation lies in the nature of modern catalog operations. Catalogers have developed a universally successful method of operating. In
theory, catalogs could be personalized, with Arthur Hughes getting a
271
272
catalog containing things that the sender assumes he would be interested
in, and Helena Hughes getting a catalog from the same company with
a different selection of products. Several catalogers have tried this. It is
technically possible. However, the increase in sales was disappointing.
Personalization of catalogs is expensive, and so few catalogers do it
because the increased sales from personalization are not sufficient to
cover the cost of personalization. Catalogers print 500,000 or more
identical copies, trying to get the per-piece cost as low as possible. They
mail these catalogs to their house file plus hundreds of thousands of
rented names. This method keeps the marketing costs to a minimum.
How to Personalize
There is a form of personalization for catalogers, however. Most catalogers produce 10 to 15 different books per year. One cataloger, for
instance, may produce some books featuring dress clothes, some with
sports clothes, and some with children’s clothes. If a customer buys
children’s clothes, she will get the children’s clothes catalogs. This
sounds like personalization. But herein lies a problem: The catalog’s
house file of previous buyers is often very small relative to the rented
list of external names. Previous buyers are always the best responders.
Why leave a previous buyer out of the sports clothing catalog mailing
just because all she has bought before was children’s clothes? It does not
make economic sense. So previous buyers will get all the catalogs.
Where does database marketing figure in that? If you cannot personalize
your communications, you cannot really do database marketing, so few
catalogers spend any money on database marketing. That is, they didn’t
until the advent of the Web.
Catalogs and the Web
My wife, Helena, buys all the clothes and gifts in our family. She is a
catalog freak. She recently spent $35,000 on catalog items in 1 year.
She does extensive research in the catalogs to find exactly what she
wants. Once she has selected the items, she usually orders by phone,
but occasionally she orders over the Web. Almost every one of the catalogs has both a toll-free number and a Web site. She rarely uses the
Web sites. She uses them only as an ordering device, once she has
found what she wants in the catalog.
This squares with Sears Canada’s experience: 92 percent of Canadian customers buying from the Sears Canada Web site have the paper
catalog in front of them. Why doesn’t Helena use the Web sites as a
shopping medium?
• She enjoys reading catalogs. They are a form of relaxation, like
reading a magazine. She does not find sitting at a computer relaxing.
• She cannot count on the colors on our PC being as true as the
colors in the catalogs
• It takes too long to find items on a Web site. She can find a
sweater or dress that she wants in a catalog in a few seconds. On
the Web, waiting for page after page to download is simply a
waste of her time.
• She can sit on the sofa and flip through catalogs. She can take
them to the kitchen or the bathroom. You can’t do any of these
things on the Web.
• Talking to an operator is much more satisfying than ordering
on the Web. Helena likes talking to people. The operators
seem to know when things are out of stock and can make
helpful suggestions for substitutes. Most Web sites don’t tell
you that the item is out of stock and very rarely make
alternative suggestions
So why does she use the Web at all? Because she hates to be put on hold
when she calls an operator. It is the result of desperation, not a choice.
More and more customers are placing catalog orders on the Web. Most
catalogers have already put all their catalog items on their Web sites, so
that customers can look at them and place orders. But the Web is not
replacing the paper catalogs. In fact, it seems to be increasing them. In
1995, 13.2 billion catalogs were mailed in the United States. By 2000,
despite the growth in Web catalog sites, the number of paper catalogs
273
274
distributed had grown to 16 billion. By 2001, about 11 percent of catalog
sales were placed through the Web.
The Web Is an Ordering Medium
The Web is an excellent ordering medium. You never get put on hold.
Each year more people learn to use the Web. Some catalogers estimate
that by 2005 they may have 35 percent of their orders placed through
the Web.
The Web, however, does not seem to be an effective sales medium.
Very few of the orders that arrive at a catalog Web site come from people
who were simply browsing the Web. What brings the visitors is the
arrival of the paper catalog. Once they had built their catalog Web site,
many catalogers tried to use advertising, mailed promotions, and affiliate
programs to bring people to the Web site to shop. However, the increase
in sales seldom paid for the cost of the promotions. The catalogers discovered that the paper catalog itself is the only really effective way to drive
customers to their Web sites. So since they still need the catalog, why do
catalogers bother to have a Web site at all? There are many valid reasons:
• Reduced ordering costs. It costs a cataloger $4 or more to take an order
over the telephone. Orders placed through the Web have a variable
cost of a few pennies. Saving $4 per order can make a big difference.
• Email communications. Catalogers have found that they can improve
the response to a catalog by using email to notify customers of its
imminent arrival. They can also use their database of previous
purchases to flag certain items for customers’ attention in an email
with a click here inside it that jumps the customer to the item in the
Web catalog. To use email communications, however, the cataloger
must have people’s email names.
• Email name capture. All catalog Web sites insist on receiving the
customer’s email name as part of the transaction. At the same time,
they permit the customer to “opt out” to receiving emails concerning
future product promotions. A vast majority of customers do not say
“No” to this question, thereby providing permission for use of their
email for commercial messages. At the same time, this permission
also involves the placement of a cookie on the customer’s computer.
• Building a relationship with customers. Before the Web, catalogers
could not afford to make a lot of nonproductive phone calls or send
direct mail to customers. They received orders and shipped them
rapidly. With the Web, this is beginning to change. When the
customer places an order on the Web, he receives an immediate
thank you on the screen, followed by a simultaneous email
confirmation. He can use the confirmation to cancel the order with
a single click. (Why would a cataloger make it so easy to cancel an
order? Because 20 percent of all catalog orders are returned. Having
the order canceled before it is shipped saves money.) When the
order is shipped, the customer gets another email: Your order was
shipped today. Finally, when the order arrives, he gets another
email: Did it arrive on time? Were you happy with it? Click here to
respond. This type of communication builds relationships like those
enjoyed by the old corner grocers who knew their customers and
talked with them frequently. This type of relationship building
increases loyalty and repeat sales. It is the first time catalogers have
been able to make effective use of a database.
• Creating one-click ordering. Amazon.com started it (as it did many
other things on the Web). You come to the amazon.com site and
find a book, video, or CD that you are interested in. You click
once, and it is shipped. There’s no need for you to give your
address, telephone number, credit card number, or other such
information. It’s like walking out of the corner grocer’s store
with a big turkey and saying, “Put it on my tab.” People love this
kind of personal life simplification and will return to a site that
makes it possible.
• Collecting demographic information. While they are taking orders on
the Web, catalogers collect demographic information from
customers. Getting customers to update their profile on the Web
is easy, and much less expensive than getting the same information
during a telephone order session. They can learn the customer’s
age, income, family composition, preferences, and lifestyle.
• Using demographic information for personalized communications. Using
this information, catalogers can send out personalized email
messages. Since catalogs are always mass produced, the
email becomes the only method of personalization.
275
276
• Web personalization. All good Web sites gain customers’ permission
to use cookies by giving them a chance to “opt out.” More than
95 percent of customers agree to their use. By using cookies, a
visitor returning to a catalog Web site sees “Welcome Back,
Susan.” Susan can see items that the cataloger thinks (or she has
told them) she is interested in based on her past purchases and
Web activity. What catalogers tried and failed to do in catalog
personalization is now economically possible using the Web.
• Control groups. Control groups can be created in the database to
test customers’ reactions to innovations in the Web site. Such
groups were almost impossible to use effectively with mass
mailings of catalogs.
• Online focus groups. In Chapter 6 we describe how Universal Music
used online focus groups to determine customers’ reaction to the
release of a new music CD. Catalogers can now use online focus
groups to determine customers’ reactions to new products.
• Measuring multibuyer performance. The Web offers the opportunity
to track what customers are looking at, which is not possible with
a paper catalog. However, few catalogers have yet taken advantage
of this feature, since it requires not just sophisticated software, but
creative marketers who can look at the data and draw profitable
conclusions from what they see. The scarcest commodity in all of
database marketing is the creative imagination needed to drive new
marketing programs. Marketers are usually awash with data that
they cannot figure out how to use profitably.
How Some Catalogers Are Using the Web
Once the landsend.com Web site was constructed, the company wanted
to try additional methods besides the paper catalog that would induce
shoppers to come to its site and buy. It introduced several new ideas that
seemed to work.
• Live customer service that allowed customers to send messages and
receive replies in real time (see Figure 12-1). The customer support
agents could synchronize their screens with customers’ so that
they could see the same screen the customer was seeing at
home. They could help shoppers with their choices and in
selecting cross sales.
• A virtual model that customers could use to try on clothes
(see Figure 12-2).
• My Personal Shopper, who helps you the way a good clerk in a
department store would (see Figure 12-3).
How well did these things work? Conversion rates among those who
used the virtual model were 26 percent higher than among those
Figure 12-1 Lands’ End Live. © Lands’
End, Inc. Used with permission.
Figure 12-2
My Virtual Model. © Lands’ End, Inc. Used with permission.
277
278
Figure 12-3 My Personal Shopper. © Lands’
End, Inc. Used with permission.
who didn’t. Conversion rates among those who used the personal
shopper were 80 percent higher than the average for other customers.
Other catalogers have also developed innovative methods to bring
customers to their site.
• Use of affiliates. Ashford.com, a jeweler, promotes its site through
12,000 affiliates who are linked to Ashford in return for a 7
percent share of the net revenue. This program brings in 15 percent
of Ashford’s sales. Ashford also uses Web advertising and Web
newsletters to previous customers.
• Web newsletters from catalogers may seem like a bonanza because
they are so inexpensive to send. However, they are also annoying.
Forrester Research found that more than half of consumers say they
receive too many commercial emails. Most long-term Net users
delete all promotional emails without reading them. However, the
value of a Web newsletter can be easily determined by creating a
control group and determining whether the total sales, number of
sales, average order size, departments shopped, and so on are greater
for those getting the newsletter than they are for the control group.
If there is no difference, drop the newsletter like a hot potato.
Problems for Catalogers Shifting to the Web
Many catalogers have fulfillment systems that use a legacy system that
is not compatible with the modern servers that are used today for the
Web and for customer service. A considerable amount of software rewriting is necessary. Only the larger catalogers can afford such an effort.
One company that succeeded in this effort was the Sharper Image
(www.sharperimage.com). The company put all its customers on a single database so that it could determine their purchases through each of
the three channels: retail stores (60 percent), catalog phone orders (23
percent), and Web orders (17 percent). The top customers were identified by lifetime value and recency of purchase. Gold customers got
special offers, preferred mailings, and gifts during the holidays. They
also got sneak-preview emails on new items.
The Sharper Image’s email list grew to over 500,000. It sent out
between two and four emails per customer per month. It also sent
out emails to new prospects, letting them know that a new catalog
was coming out and offering a merchandise certificate. The emails did
not reduce the number of paper catalogs sent out; they increased it.
The number of catalogs sent out by the Sharper Image grew by 20 percent every year, with 62 million being sent out by 2000. Both catalogs
and special mailers worked for this company.
The Web Creates Multibuyers
A survey by the National Retail Federation showed that catalog customers who buy on the Web buy 20 percent more than customers who
respond to a catalog only by mail or phone. Web shoppers also spend
33 percent more in retail stores than retail shoppers who do not use the
Web. A survey of 48,000 shoppers in all channels by Shop.org found
that shoppers who bought via all three channels—retail store, catalog,
and the Web—represented 34 percent of all Web shoppers. In-store
shoppers who also bought from the same retailer online spent an average
of $600 more per year than typical in-store shoppers.
Macys.com saw its Web site as a stand-alone store carrying items that
were not available in its catalog. Customers visiting the Web site registered their age, gender, family status, and the product lines of greatest
interest to them. Macy’s learned whether the customer planned to buy
a home, take a vacation, have a baby, or get married anytime soon.
279
280
Macy’s click-through rates on emails featuring items on the Web site
averaged 10 percent, with 2 to 3 percent of those clicking making a purchase. When customers make a purchase at the macys.com site, a pop-up
window asks them to rate their Web experience. The company found
that many customers left the site quickly if they did not find what they
wanted in the search window. As a result, when a search failed, Macy’s
visitors got a window showing the address, phone number, and directions to the nearest Macy’s store.
Office Depot introduced Web ordering for its catalog as Office
Depot Online. It has been profitable since day one. The company soon
discovered that catalog customers who ordered through the Web spent
33 percent more than other catalog shoppers. In 2001, Office Depot
Online sales grew by 30 percent to $1.5 billion, representing 14 percent of all Office Depot sales. Office Depot, like Dell, set up Web pages
for special customers such as Bank of America. By 2001, 85 percent of
Bank of America’s office supplies were provided by Office Depot
through the Web. These Web pages permitted Office Depot to make
special volume price adjustments for large customers, which it could not
easily arrange without the Web to keep track of the hundreds of thousands of individual orders.
Use of Collaborative Filtering
The Internet has opened up a range of new opportunities for personalization that may lead many catalogers to reconsider their attitude
toward database marketing. Amazon has shown us the way. Amazon
uses emails and its Web site to suggest books and movies to customers
that they might be interested in, using collaborative filtering to come
up with the suggestions.
Collaborative filtering is provided by www.Netperceptions.com,
among others. It consists of software that makes use of a client’s database of customers and their previous purchases. The goal of the software
is to determine what a person’s lifestyle and preferences are and then to
find other customers within the database who have similar tastes. Whatever the cataloger had sold to those similar customers might appeal to
you, so the cataloger suggests it to you. One of the most convincing
uses of collaborative filtering was an experiment tried with Great Universal Stores (GUS), the largest cataloger in the United Kingdom. Before
collaborative filtering, GUS customer service reps were getting a 20
percent cross-selling rate on orders placed from the paper catalog. That
means that in one call out of five, the rep was able to sell the customer
an item from the catalog that the customer had not intended to buy.
Netperceptions.com introduced collaborative filtering into the GUS
call center software, using the huge GUS customer database. The suggested next best product showed up on the customer service rep’s screen
as soon as the rep typed in the first purchase requested by the caller. The
company’s previous success at cross-selling was due to the customer
service rep’s native intelligence. If a woman ordered a dress, the rep
might suggest matching shoes, a belt, or a handbag. The suggestions
from the collaborative software were described by the GUS reps as
“spooky.” The software might suggest selling towels to the woman who
ordered the dress. Towels? What was the relationship of towels to the
dress? There was none—except that the software had found similar
women in the GUS database and had found that many of them also
bought towels. The amazing thing about the software was that it
worked. After getting used to it for 6 weeks, GUS customer service reps
were getting a 40 percent cross-selling rate—double what they had been
getting before. Now that is database marketing, folks. Collaborative
filtering is used by more than 200 American companies. It requires a
large database and super fast software. It is not for the mom-and-pop
catalog operation, or for the timid.
In the past 2 years, I have probably spent more than a thousand dollars with Amazon, some of which was the result of Amazon’s collaborative filtering suggestions. One email that I received a year ago has
stuck in my mind. It said:
Two years ago you bought Dark Sun by Richard Rhodes. You may be
interested to learn that Richard Rhodes has just had a new book published called Why They Kill. To learn more about this book, or to
order it, click here.
281
282
Well, I clicked, I bought it, and I loved it. This new book had nothing to do with the previous book except the author. This is database
marketing.
What Catalogers Have to Guard against When
They Use the Web
• Web search for products is slow. Flipping through a catalog is easy. You
can find what you want and order it. You can’t flip through a Web
site. To find a cardigan sweater in Appleseed’s catalog took about
45 seconds. To find the same sweater from scratch on Appleseed’s
Web site took 3 minutes. To find the sweater on the Web site
using the catalog number took about 1 minute. Fast search is
possible, but only if very clever marketers team up with very
clever Web programmers.
• The colors on the Web are not true. There is no way that catalogers
can know what users’ monitors are doing to the pictures of their
products. There is no known cure for this disease.
• Out-of-stock information is seldom accurate. When you call a
telemarketer, her software tells her and she tells you that the item
you want is out of stock. Web sites seldom do. As a result, you
place the order and find out by email a few days later that the
item is out of stock. After a few experiences like this, you stick
with a live operator. This is really inexcusable. To get customer
acceptance, Web sites have to be as good as or better than a live
operator. They have to know what is out of stock in real time and
be able make instant recommendations on substitutes.
Lifetime Value Changes for Catalogers
Let’s see what would happen to a cataloger if it were to adopt the database initiatives proposed in this chapter. To understand the change, let’s
start with a traditional successful cataloger. This company has already
put its catalog on the Web. It sends out 2,000,000 books every 6
months, at an in-the-mail cost of $2 per book. The company gets an
overall response rate of 4 percent. Included in the mailing are 500,000
new names. These produce responses from 20,000 first-time buyers.
We are going to trace the lifetime value of these 20,000 first-time
buyers over four rounds (2 years). The average order size on the first buy
is $120, of which the product and fulfillment costs are 40 percent. In
the second-round mailing 6 months later to these 20,000 first-time
buyers, the purchase rate is 40 percent and the average order size is
$130. This rises over the next year to $150, and the purchase rate rises
to 50 percent. Of the original 20,000 first-time buyers in round 1,
there are 1800 who are still buying 2 years later. Meanwhile, of course,
the company has proceeded with additional prospect mailings that are
producing identical charts every 6 months. The lifetime value of the
original 20,000 first-time buyers after the fourth round is $67.68, as
Let us now assume that the cataloger has adopted many of the recommendations made in this chapter. The company has spent $1,200,000
on software for its Web site that does several significant things:
• Asks for customer profiles, capturing email names and customer
preferences and demographics. This reaches 20 percent on the
first round and an additional 20 percent on each additional round.
Table 12-1 LTV of Catalog First-Time Buyers
Round 1
Round 2
Customers
Purchase rate
Average purchase
Revenue
20,000
100%
$120
$2,400,000
8,000
40%
$130
$1,040,000
3,600
45%
$140
$504,000
1,800
50%
$150
$270,000
Cost percent
Costs
Number mailed
Mailed cost
Mail costs
Total costs
40%
$960,000
500,000
$2.00
$1,000,000
$1,960,000
40%
$416,000
20,000
$2.00
$40,000
$456,000
40%
$201,600
20,000
$2.00
$40,000
$241,600
40%
$108,000
20,000
$2.00
$40,000
$148,000
$440,000
1.00
$440,000
$440,000
$22.00
$584,000
1.04
$561,954
$1,001,954
$50.10
$262,400
1.08
$242,963
$1,244,917
$62.25
$122,000
1.12
$108,699
$1,353,616
$67.68
Profit
Discount rate
NPV of profit
Lifetime value
Round 3
Round 4
283
284
• Provides one-click ordering for all customers who complete the
profile, along with personalized emails and Web site opening pages.
• Sends an email thank you for each order, an email when the order
is shipped, and an email asking if the order was satisfactory.
• Provides personalized emails announcing the arrival of each
book, with click here for items that it is assumed the customer
is interested in.
• Uses collaborative filtering to make suggestions on next best
product, leading to a cross-selling rate of 40 percent.
• Offers an advanced search feature that permits customers to find
products faster.
• Has live shopper support for people who want to have a text
chat with a customer service rep while they are on the site, or
who want to call on another line to talk to a live agent.
The $1,200,000 software, amortized across 3 years, will represent
a cost of $200,000 for each group of first-time buyers acquired
($1,200,000/3 ⫽ $400,000/2 per year ⫽ $200,000).
The result of these changes can be seen in the Table 12-2.
The purchase rate for the second-time buyers has risen from 40 percent to 50 percent. The average order size has increased. The costs have
Table 12-2 Improved LTV with Web Site Software
Round 1
Round 2
Customers
Purchase rate
Average purchase
Revenue
20,000
100%
$120
$2,400,000
10,000
50%
$140
$1,400,000
5,500
55%
$155
$852,500
3,300
60%
$170
$561,000
Cost percent
Costs
Number mailed
Mailed cost
Mail costs
Software cost
Total costs
40%
$960,000
500,000
$2.00
$1,000,000
$200,000
$2,160,000
38%
$532,000
20,000
$2.00
$40,000
$0
$572,000
37%
$315,425
20,000
$2.00
$40,000
$0
$355,425
36%
$201,960
20,000
$2.00
$40,000
$0
$241,960
$240,000
1.00
$240,000
$240,000
$12.00
$828,000
1.04
$796,743
$1,036,743
$51.84
$497,075
1.08
$460,255
$1,496,998
$74.85
$319,040
1.12
$284,256
$1,781,254
$89.06
Profit
Discount rate
NPV of profit
Lifetime value
Round 3
Round 4
Table 12-3 Gains from Software Improvements
Round 1
New LTV
$12.00
Old LTV
$22.00
Difference
($10.00)
Times 80,000 ($800,000)
Round 2
Round 3
Round 4
$51.84
$50.10
$ 1.74
$139,156
$74.85
$62.25
$12.60
$1,008,323
$89.06
$67.68
$21.38
$1,710,552
gone down as a result of increased use of the Web rather than telephone
calls. Against that, there is a software cost for each group of first-time
buyers of $200,000. Lifetime value has gone up to $89.06, from $67.68.
Overall, the changes will increase profit by a total of $1.7 million
over 2 years, as shown in Table 12-3. Note that these profits of $1.7
million include all costs, including the heav y losses in round 1 that
resulted from the initial software costs.
Each cataloger will have to work out its own figures based on its own
situation. In the end, however, the results will probably be close to those
shown here. When it is done right, database marketing works. Customer
communications work. People respond when people they know and have
done business with write to them and pay attention to them. Statistics
from other catalogers discussed in this chapter show that when you
adopt the improvements listed in this chapter
•
•
•
•
The repurchase rate will go up.
The average order size will go up.
The cost of order fulfillment will come down.
Cross sales will go up.
Building B2B Catalog Sales
A business-to-business cataloger with sales in excess of $14 million had
never built a marketing database. It turned to Sejal Shah at CSC Advanced
Database Solutions to come up with a plan. Here is what he found:
• The cataloger had about 55,000 current customers who spent an
average of $250 per year.
• It mailed about 300,000 catalogs every quarter at $1 each and had
an annual response rate of 6.2 percent.
285
286
• Its margin on product was about 53 percent, out of which it had to
take the cost of the catalogs and about 10 percent for fulfillment.
Sejal recommended two methods for building sales with a database:
1. Building a customer database and using it to determine the
profiles of the most profitable customers, then using those profiles
to mail catalogs to companies whose profiles matched the profiles
of these profitable customers
2. Using emails to communicate with customers, leading to increased
cross sales, larger order size, increased frequency of orders, and
increased retention
Creating Customer Profiles
The method that Sejal proposed was this:
• Once the database was built, append D&B data and use analytical
software to determine the profile of the most profitable customers.
Each customer profile would include these data:
• SIC codes
• Total purchases in the past 12 months
• Key product groups purchased
• Total employees and total sales
• Using this analysis, marketers at the cataloger could develop a
market penetration report for each profile. The penetration report
would look something like Table 12-4.
Table 12-4 Market Penetration
Profile
group
Company
customers
U.S.
universe
Market
penetration
A
B
C
D
E
F
Total
10,221
8,874
3,341
2,116
1,984
445
26,981
109,334
99,447
45,331
69,004
83,556
4,801
411,473
9.35%
8.92%
7.37%
3.07%
2.37%
9.27%
6.56%
This report clearly identified those companies that were the most
likely prospects for further sales. Using these profiles, the company
could acquire the names of prospects that matched the SIC codes and
other data in the profiles.
Using these numbers as a base, what results could the cataloger
expect from the prospect mailings? Certainly not the 6.2 percent
received on the existing catalogs. Those were sales to existing customers,
many of whom had been purchasing from this cataloger for years. Sejal
estimated that with the new mailings, the cataloger would get about
one-third of the sales rates that it got from the regular catalog mailing.
As a result, he predicted that the cataloger would get a response in the
first year from about 2 percent of those who received a catalog.
How many catalogs should be mailed? Sejal proposed three scenarios:
a conservative, a bold, and an aggressive one. In the conservative scenario, the cataloger would mail to about 150,000 new names each
quarter, or 600,000 during the year. The bold total would be 1 million
per year, and the aggressive would be 1.6 million in total.
Putting everything together, Table 12-5 shows what Sejal projected
for the results of the first year’s increased mailings. As a result of the
new database and mailing strategy, the cataloger would gain between
$3 and $7 million in sales, spend $200,000 for the database and software, and still make a substantial profit. Many companies expect to lose
Table 12-5 First-Year Results
New mailing
Response rate
New customers
Average order
Revenue
Name rental ($100/M)
Catalogs @ $1
Fulfillment @ 10%
Database
One-time DB costs
Cost of goods (47%)
Total costs
Net profit
Investment
Conservative
Bold
Aggressive
600,000
2.20%
13,200
$250
$3,300,000
$60,000
$600,000
$330,000
$200,000
$140,000
$1,551,000
$2,881,000
$419,000
$1,000,000
1.42
1,000,000
2.00%
20,000
$250
$5,000,000
$100,000
$1,000,000
$500,000
$200,000
$140,000
$2,350,000
$4,290,000
$710,000
$1,440,000
1.49
1,600,000
1.80%
28,800
$250
$7,200,000
$160,000
$1,600,000
$720,000
$200,000
$140,000
$3,384,000
$6,204,000
$996,000
$2,100,000
1.47
287
288
money in their first year of database marketing and recoup it later. This
study shows that for this company, it was not necessary to predict a loss.
Furthermore, this chart includes only new sales. Sejal assumed that
the cataloger would also continue to get the existing annual sales of
about $14 million.
One-Time Setup Costs
In addition to the annual database costs, there were one-time setup
costs, shown in Table 12-6.
Email Marketing
Sejal projected that in the second year, the cataloger could gain from
email marketing to its existing customers. The use of email marketing was possible only when the cataloger had a substantial number of
customer records that included email names and permission to use
them. Since the cataloger did not have many email names to start with,
Sejal suggested that it become aggressive about capturing email names
on all future sales, thus rapidly building up a base of customer email
names.
What could be done with email marketing? Sejal suggested several
techniques:
• Regular promotions of existing products, particularly before and
after the catalog arrives
• Promotions of sale items
• Last-minute specials
• Retention messages (thank you for your business, customer
surveys, etc.)
• Follow-up messages (thank you for your order; your order will be
shipped on . . .)
Table 12-6 One-Time Setup Costs
Needs assessment
Design and development
Training
Total
$
$
$
$
30,000
100,000
10,000
140,000
• Viral marketing (“New product specs are available. Send them to
a coworker.”)
• Email newsletters with product information
What results might the cataloger get from email marketing? The
experience of other catalogers suggests a lift in sales of between 5 and
20 percent using well-targeted and appreciated emails. For this cataloger, Sejal conservatively projected only 5 percent. He used that figure
to estimate the increased sales and profits from emails (Table 12-7).
By adding the email profits to the prospecting profits, Sejal was
able to predict the cataloger’s profits for the second year, as shown in
Table 12-8. The second year’s response rate was projected to increase
Table 12-7 Profit from Email Marketing, Year 2
Conservative
Bold
Aggressive
10,000
18,600
28,600
$7,150,000
$357,500
$125,125
10,000
30,000
40,000
$10,000,000
$500,000
$175,000
10,000
46,400
56,400
$14,100,000
$705,000
$246,750
Old customers
New customers
Total
Sales @ $250
5% increase
Profit @ 35%
Table 12-8 Second-Year Results
Conservative
Bold
Aggressive
New mailing
Response rate
Customers
Average order
Revenue
600,000
3.10%
18,600
$250
$4,650,000
1,000,000
3.00%
30,000
$250
$7,500,000
1,600,000
2.90%
46,400
$250
$11,600,000
Name rental ($100/M)
Catalogs @ $1
Fulfillment @ 10%
Database
Cost of goods (47%)
Total costs
$ 60,000
$ 600,000
$ 465,000
$ 350,000
$2,185,500
$3,660,500
$ 100,000
$1,000,000
$ 750,000
$ 350,000
$3,525,000
$5,725,000
$ 160,000
$1,600,000
$1,160,000
$ 350,000
$5,452,000
$8,722,000
New mail profit
Email profit
Total profit
$ 989,500
$ 125,125
$1,114,625
$1,775,000
$ 175,000
$1,950,000
$2,878,000
$ 246,750
$3,124,750
Investment
$1,010,000
2.10
$1,450,000
2.34
$2,110,000
2.48
289
290
from 2 percent to 3 percent for a number of reasons. The first year’s
mailing would show which SICs were pulling and which were not, so
the cataloger should get much better at selecting names. In addition,
many of the customers mailed to in the first year would be getting their
second year’s worth of catalogs. Since the long-term projection of catalog
response for this cataloger was 6 percent, these second-year customers
should respond at a rate closer to that, bringing the average up to 3 percent or higher. The only surprise in the second year is the increased
cost of the database. The software used in the first year did not include
a provision for email marketing. Sejal recommended a software upgrade
that moved the database cost up to $350,000 per year.
What this case study shows is how to go about calculating the benefits
from database marketing. Too many people think that it is a nice concept,
but have no idea of how to prove it.
Summary
Catalogers never made much use of database marketing until the
advent of the Web, because the nature of modern cataloging made it
uneconomical to provide personalized communications. Now that the
Web has arrived, catalogers should rapidly begin to use database marketing to build relationships with customers and to increase loyalty
and cross sales. The methods are out there and are within the reach
of all but the smallest catalogers. The increases in sales and profits can
be substantial.
What Works
• One-click ordering
• Personalization of the Web site with cookies
• Cross sales accompanied by personalization and collaborative
filtering
• Using the Web site to capture customer profiles, email names,
and preferences
• Saving order fulfillment costs through increased use of the Web
• Email communications to drive traffic to the catalog and the Web
• Using customer-created profiles to capture email names and
demographic information
• Using profiles to create personalized communications and
Web sites
• Using control groups to measure the response to customer
communications
• Online focus groups using captured customer email names
• Measuring multibuyer performance and using it to drive
marketing strategy
• Live customer service on a catalog Web site
• Virtual models created by the customer
What Doesn’t Work
• Expecting to drive traffic to a catalog Web site without a paper
catalog
• Fancy flash pages that take too long to download
• Web sites that lack the information known to telephone sales reps
• Web newsletters, which may be overused
• Slow product search on Web catalog sites
Quiz
1. Catalogs are succeeding today for all but which of the following
reasons?
a. Personalized catalogs
b. Lightning-fast fulfillment
c. Universal credit cards
d. Sophisticated telesales
e. No-hassle returns
2. Why don’t many catalogers today do database marketing?
a. It provides no sales lift.
b. They have never heard of it.
c. The economics of catalog production.
291
292
3.
4.
5.
6.
7.
8.
9.
10.
d. All of the above.
Give three reasons why catalogers have profitable Web sites today.
What did Office Depot find was the valuable feature of customers
who shopped both at the retail stores and on the Web?
When GUS adopted collaborative filtering, what marketing
method did it change and by how much?
a. Catalog distribution was down 8 percent.
b. Cross sales were up 100 percent.
c. Web visits were up 15 percent.
d. Call abandonment was down 10 percent.
List four things that catalogers need to add to their Web sites if
they are going to succeed on the Web.
What did Macy’s do when it found that customers were leaving
its site when they didn’t find what they wanted?
What is the advantage to Lands’ End of having a personal
shopper? Conversion rates were up:
a. 10 percent
b. 20 percent
c. 30 percent
d. 40 percent
As the Sharper Image email campaign grew in size, what
happened to its catalog distribution?
a. It went down 5 percent
b. It went down 10 percent
c. It went down 20 percent
d. It went down 25 percent
What is the advantage of a Virtual Model to Lands’ End?
Conversion rates were up by
a. 16 percent
b. 26 percent
c. 36 percent
d. 46 percent
e. 56 percent
C
H A P T E R
13
Finding Loyal
Customers
For years, retailers have argued that having regularly advertised, deeply
discounted prices brings price-oriented customers into their stores but
that, over time, these customers convert to regular, profitable customers.
Research done by the Retail Strategy Center Inc. based in Greenville,
South Carolina, shows that this widely held belief is a myth=?
. A handful of these customers do convert into “good” regular customers, but the majority actually defect within twelve months of their first
shopping visit. I have yet to find a retailer anywhere in the world whose
investment in this type of shopper has yielded an attractive return on
investment.
Brian Woolf, Customer Specific Marketing
M
ost companies are set up for customer acquisition. Few are set up
for customer retention. Yet, as you can tell from reading this
book, retention is almost always more profitable than acquisition. Most
companies lose money on the first sale to a customer. The profits they
make are derived from the second and subsequent sales. Does that mean
that companies can ignore acquisition? No. It means that acquisition
should be conducted with long-term retention in mind.
Frederick Reichheld, in his book The Loyalty Effect, which I cannot
praise too highly, points out that the method a company uses to acquire
customers often determines whether the company will be successful in
the long run. Deep discounts, for example, do not create loyalty. They
cheapen the image of the brand, cost you valuable margin, and focus
293
294
the customer’s attention on the wrong thing: the price of the product
instead of its quality and the service you offer.
Finally, we cannot assume that every acquired customer is good for
the firm. Fleet Bank discovered that half the customers it was acquiring
destroyed value and would never be profitable for the bank.
There are dozens of acquisition methods, several of which are covered in case studies in this chapter. In general we can summarize them
as follows:
• Mass marketing through TV, radio, and print ads. With generaluse products and packaged goods, mass marketing is far preferable
to direct marketing for customer acquisition. Of course, to acquire
customers through mass marketing, you have to have a method of
capturing customers’ names, addresses, and email names when
they buy in retail stores. Without this step, you have sold product,
but you have not acquired a customer.
• Creating profiles of your existing customers, then renting names of
prospects who resemble your profitable customers.
• Exchanging customer names with other businesses. The catalog
industry uses this idea extensively. It works very well. This is
contrary to the idea that is prevalent in the rest of the world,
particularly Europe and Japan, that customer names are a
corporate asset that could be ruined if the company were to
allow them to become known to any outsider. Renting customer
names has become a profit center in many companies and, as
such, is a positive good to the direct marketing industry.
• Providing a method by which customers can register with you when
they buy a product. This is done for computers and software and
for many electrical products. Millions of people willingly register.
The best method in use today for getting people to register is
through a Web microsite that you set up specifically for this
purpose. The site should be listed on the product packaging or on
the product itself. Providing a reward helps.
• Determining which clusters your consumer customers come from and
seeking additional customers in other households located in
those clusters.
F i n d i n g L o ya l C u s t o m e r s
• Doing penetration analysis by zip code or SIC code and finding out
where you are strong and where you are weak. Armed with this
knowledge, you can better direct your acquisition strategy. It is
easier to sell in areas where your name is a household word than
in areas where you are unknown.
• Giving customers an incentive to provide you with referrals. As we will
show you, referred customers are better than the average customer
and are worth cultivating.
Penetration Analysis
Table 13-1 gives a simplified example of a penetration study for a company that sells to several industries.
The company has 29,000 customers, roughly divided into four sectors.
As you can see, most of the company’s sales come from light manufacturing. The companies with the highest lifetime value are in heav y
manufacturing. But the high-technology sector is particularly interesting. The U.S. universe comes from Dun & Bradstreet data. It shows the
number of companies in the SIC codes that Weldon sells to. As you can
see, Weldon’s penetration in the high-technology sector is very high: 42
percent. This means that Weldon is a household word in this industry.
It will be much easier for Weldon to pick up an additional customer in
this sector than in any of the other three because it is so well known
there. This shows the power of penetration analysis.
Table 13-1 Penetration Analysis
Weldon
U.S.
customers universe
Metal
production
Light
manufacturing
Heavy
manufacturing
High
technology
Total
Penetration
ratio
Lifetime
value
Target
potential
Rank
$15,144.68
3
$563.70
4
254
2,433
10.44%
$145,067
15,442
162,009
9.53%
$5,914
44
1,288
3.42%
$988,145
$33,756.51
2
612
29,198
1,453
167,183
42.12%
17.46%
$127,675
$53,776.39
1
295
296
Target potential is derived by simply multiplying the penetration
ratio by the lifetime value. This is a simple approach, but it has value.
Here is another approach to customer acquisition for a business-tobusiness situation.
Finding Leads for Equipment Financing
Equipment financing is not really like any other type of business. The
amounts involved can be huge. The terms, the length of the lease, and
the items financed are seldom standard (as they are in automobile or
home mortgage financing). It is hard to predict companies’ need for
financing, as decisions on equipment purchases are seldom made on a
predictable schedule. As a result, finding leads is tough. Which companies are going to need financing, and when?
The situation is complicated by the fact that the industry is highly
competitive. In addition to banks, there are finance companies that specialize in equipment financing. There is very little customer loyalty, as
decisions are made primarily on rates rather than on provider affinity,
service, or brand. To further complicate the picture, the equipment
financing sales force, like all sales forces, is highly skeptical of central
assistance. The salespeople tend to be independent and are more comfortable developing their own sales processes than participating in a
group effort. To grow its equipment financing business, KeyBank’s
direct marketing group had to come up with a system that served up
leads to its Key Equipment Financing sales force in a way that was both
productive and acceptable to the staff.
Key’s direct marketing group developed a unique way of solving this
problem with the help of a team from SIGMA Marketing Group. The
solution involved a combination of direct mail, telemarketing, and leasing managers in a system driven by a model using Dun & Bradstreet data.
In analyzing the situation, both groups saw that Key’s 30 leasing
managers were spending entirely too much time trying to find leads.
The life of a leasing manager was very demanding, with constant travel
and much time spent chasing possible leads that might not materialize.
Once they had a solid lead, the leasing managers found that closing the
sale could be a fairly straightforward process. SIGMA recommended
setting up a lead-discovery process that would produce qualified leads
for the leasing managers so that they could focus on closing sales and
not waste time on lead development.
There were four parts to the system:
1. Create a relevant and highly targeted prospecting list that the
leasing managers would find credible.
2. Develop a compelling direct-mail communication on behalf of
each leasing manager that would permit effective telemarketing
follow-up.
3. Have outbound telemarketing associates work closely with the
leasing managers so that the leads and appointments generated
would be relevant and worthwhile.
4. Create a lead-generation database to track and identify the
business generated by the leasing managers using this program
in terms of both pending and closed deals.
Creating a Universe of Prospects
Using Key’s data, SIGMA created a CHAID (chi-square Automatic
Interaction Detector, a statistical technique) model to identify the characteristics of Key’s current leasing customers. This model identified the
best predictors of a company’s likelihood to be a leasing customer. They
were (in order)
•
•
•
•
Company sales volume
Year company started
Company number of employees
Dun’s Major Industry
Ten segments were created. The top two segments, with ownership
indices of 269 or greater, were
1. Ownership rate of 30.5 percent, with Dun’s Major Industry
classification of Mining, Communications/Utilities, Business
297
298
Services, or missing and year company started being pre-1960
or missing.
2. Ownership rate of 20.2 percent, with Dun’s Major Industry
classifications of Mining, Communications/Utilities, Business
Services, or missing; year company started being 1960 to the
present; and sales volume greater than $5 million.
A sign-up form was developed for each leasing manager that captured
not only all their contact information, but also those attributes that they
wanted a lead to have, such as geography, sales size, and industry. Each
leasing manager furnished a sample of his or her signature for use in
the outgoing communications.
The information from these two sources was combined into a list
order for Dun & Bradstreet. The D&B list was divided geographically
by leasing manager, and the leasing managers were asked to make any
additions, edits, or deletions to the list that they wanted.
Armed with a good, edited list, Key and SIGMA worked to develop
compelling and relevant direct-mail communications that could be
mailed on behalf of each leasing manager. The leasing managers were
able to schedule the mailings by geography and quantity over 2-week
intervals, staggered so that they would not have more leads than they
could handle at any one time.
To set up appointments, one outbound phone associate was dedicated
to each group of three leasing managers. The associate held phone conversations with each leasing manager on a regular basis to understand
the specific nuances of what each leasing manager valued in a lead (e.g.,
the prospect’s role, budget, or timing). Some of the leasing managers
even wanted the associates to set up their appointments for them and
provided the associate with their calendars. The leasing managers could
use email or a toll-free number to talk to their associate while they
were in the field on visits. When the associate generated a lead or set
up an appointment, the associate would call and email the facts and
details of the lead (e.g., who, what, when, and desired amount) to the
leasing manager. Upon receiving the message, the leasing manager
would often call the associate back to obtain further information
about the prospect (e.g., personality, potential barriers to acceptance,
and who the decision maker was).
Of course, many prospects were not ready to do business at the time
the calls were made. The associate would then ask when they might
need equipment leasing in the future and would note the answer in the
calling database. Software in the calling database automatically brought
the prospect’s name back to the associate’s attention when the time for
a follow-up call had arrived. As the callers gained valuable information
from the initial phone contacts, the information was added to the calling
database. As a result, the lead rates steadily increased over time as the
aggregated knowledge from the calling database grew.
Finally, a series of daily and weekly reports was created to measure
mail, calling, and lead activity overall. Additional reports were created
to track mail, calls, and leads at the leasing manager level so that best
practices could be identified and shared with the rest of the group.
Measuring Success
How do you know if a program like this is really working? It may look
good on paper and it may seem to be generating leads and appointments, but is it really better than the old system of letting leasing
managers forage for themselves? How can you prove that the money
spent is really worthwhile? There is a really simple way to determine
whether a program like this is successful: Are the leasing managers
using it? At first, only 7 of the 30 leasing managers signed on to use the
system. As the success of the program became clear, more and more
leasing managers signed on. At the end of 6 months, 20 of the leasing
managers were actively using the system. It became the central focus of
the Key Equipment Financing lead-generation program. Within 4
months, the lease accounts generated by the program as a percentage
of total accounts grew from 1 percent to 1.5 percent—an increase of
50 percent. At the same time, the program also boosted the loan
accounts generated by the program from 2 percent to 4 percent—a gain
of 100 percent in 4 months.
299
300
Managing Leads on the Web
One company provided content management solutions to more than
1200 companies worldwide. In its marketing programs, the firm produced online events, conducted telemarketing, and sent out direct mail
and email. Despite the advanced technology of its products, it did not
have very good systems for lead management. A review of its situation
disclosed that the company had
•
•
•
•
•
No lead-management system or method of lead-quality measurement
No system for lead reporting, distribution, or tracking
No e-marketing campaigns
No integration of outsourced telemarketing with other lead activities
An average of two online events per month that produced about
300 leads per quarter
• An average cost per lead that exceeded $500—more than double
that of other programs
To solve these problems, the company created a new, home-grown
lead-management solution. The core component was a central database
that cleaned and standardized information on leads for distribution to
the corporate sales force automation tool, SalesLogix. The system provided a closed-loop tracking mechanism that assigned a unique ID to
each lead. The system included
• An online event registration system that captured not only the
contact information, but also data about the lead and its possible
need for the company’s products.
• A global campaign information database that was used by
marketing managers to input campaign details through a Web
interface. The information was sent to the SalesLogix system to
track the lead distribution resulting from the campaigns.
• The lead-submission process was changed from sending random
emails, Excel spreadsheets, or voice messages manually to the use
of an online Web form. This Web form allowed standardization of
the information that was gathered about contacts.
• Marketing activities conducted outside the Web, including
telemarketing activities, were transferred to the new system.
The three main functions of the new system were
1. Importing. The system used the email address as a key. When this
was not available, it constructed a match key from the first and
last names of the contact and the company name. A log was kept
of all data imported to or exported from the database.
2. Processing. Country names, state names, area codes, phone numbers,
zip codes, region codes, and activity dates were standardized.
Competitors’ names were removed from the database. Standard
SIC codes were appended. Leads were assigned to the appropriate
salesperson. Incomplete records were flagged, and blank fields were
populated, either by software or manually. Separate tables were set
up for individual contacts, activity information, event details, and
data sources.
3. Reporting. Leads were furnished to the SalesLogix system daily.
Daily, weekly, and monthly reports showed the growth of the
database in the past, as well as the current volume of leads
in the database. The system also created a weekly partner
report to show the leads generated by partners.
Results
The new system decreased the cost per lead to about $50. It did so
through
• Establishing a 100 percent tracking mechanism for every lead that
was generated worldwide
• Setting up lead-quality metrics that filtered out 25 percent of the
raw leads
• Increasing lead volume from 300 to 9600 per quarter
• Increasing online events from 2 per month to 12 per month
• Introducing biweekly e-marketing campaigns for customer and
partner communications
301
302
• Bringing telemarketing in-house and integrating it with the leadflow process
• Setting up a global lead-distribution and lead-reporting system
that operated through the Web
Building a Relationship with Prospects
BMW developed a unique program for building a relationship with
prospects even before the prospect made a firm decision to purchase a
car. The process was called “Extending the Dialogue.” BMW produces
luxury cars that have a fierce customer brand loyalty. At the time that
BMW started this program, the entire luxury car marketplace was
sharply down—sales were 9 percent below the previous year’s sales. In
addition, BMW’s key competitors, Mercedes Benz and Lexus, were
expanding their product lines with strong product offerings.
The start of the new BMW program was a real-time prospect scoring system that identified BMW’s “best prospects.” The program was set
up to communicate with these targeted prospects over a 6-week period.
Before the program could begin, Kay Madati, director of relationship marketing, the genius behind the program for BMW, asked
msdbm of Los Angeles for help in overcoming a number of significant obstacles:
• Twelve different data systems had to be combined into a common
marketing database.
• BMW had to get multiple departments to cooperate with one
another and to share their data freely with a central source.
• BMW had to create an environment in which there was unselfish
sharing of response and sales information among competing vendors.
• The company had to get a highly successful and independent
dealer organization to share its information with BMW.
• It had to get prospects to describe their purchase intentions.
• It had to recognize that high-end consumers do not respond
well to direct marketing efforts because they receive them from
many companies.
• It had to deal with the fact that many consumers’ computer
equipment had slow Internet connections and Web browsers
(like AOL) that could not receive HTML emails.
• BMW had not yet computed the lifetime value of its customers.
• At first, BMW could not deliver personalized content about the
products that the customers owned or were interested in.
• BMW lacked a scoring system to determine who would be a great
or a valuable prospect.
• In the past, BMW had not used accurate methods of evaluating
success, such as control groups.
• BMW had to shift from being an entirely product-oriented
company to being a customer-oriented company.
How the System Worked
The system that msdbm designed for BMW worked this way:
• First, BMW and msdbm profiled and analyzed current owners to
determine who would make the best prospects, relying on both
customer-supplied and appended data from 13 different sources.
• They decided to concentrate marketing dollars on these best
prospects, using a 6-week communication program that involved
both direct mail and email.
• They created lifetime value tables for the owners of new and
used cars.
• They got a strong manager to work with the various vendors,
internal departments, and dealers to encourage them to share
information.
• They developed a process and methodology for a true relationshipbuilding system using not only the Internet but also data from
BMW’s legacy systems, changing and updating all systems so that
they could receive and use responses from the prospects.
• They surveyed participants to receive feedback on the program,
refine the process, and improve the delivery system.
• They set up a control group to determine the effectiveness of
the program.
303
304
• They provided a Web-based reporting system to track the results.
• The system was set up to manage 5 million prospects and
customers in the database.
• They provided feedback on all the information to everyone in real
time through the Web.
The Scoring System
Before prospects were put into the system, they had to be scored to
identify those who were more likely to purchase, by model, and those
who were less likely to purchase. The scoring system set up by msdbm
automatically
• Performed regression and correlation analysis using customersupplied and Experian data on current BMW owners to identify
unique independent variables.
• Developed score ranges for variables based on frequency distribution,
standard deviation, and historical experience.
• Scored samples of prospects and compared results.
• Developed lifetime value tables and assigned a value to each customer
and participant.
• Delivered personalized content and information to each responder,
whether the responder was a previous customer or a prospect.
• Used variables to score prospects; these variables included
purchase time frame, make of current vehicle, age, income-toprice ratio, year of current car, source of prospect lead, and debtto-income ratio.
BMW and msdbm set up a 6-week communication program that was
targeted to people who had expressed an interest in BMW by responding to one of the current marketing programs, such as direct-mail or
Internet requests. These prospects had come from rented mailing lists,
in-house lists, and responses to advertising. Although this was a prospectbased program, current BMW customers who were interested in purchasing another BMW and who had responded to existing BMW
programs were included.
The media used were print on demand, email, and Internet-based. The
content was designed to be what the prospect was interested in receiving.
There were a number of successive follow-ups that combined dealer calls,
emails, personalized on-demand printing, and invitations to events.
Results of the Program
The effectiveness of the program was gauged by comparing the conversion rate of consumers enrolled in Extending the Dialogue to that
of a control group. Conversions by participants in the program were
20 percent higher than in the control group. In detail, the program had
these results:
• Communications were sent out to prospects and customers in a
much more organized and timely way.
• The system beat the control group by 150 percent in terms of
closing the sale.
• Sales of BMWs were up 25 percent for the year, while sales of all
other luxury imports were down 7 percent.
• Overall sales increased by $25,000,000.
• Both management and dealers felt that they had a better
understanding of the BMW customer, which allowed BMW to
better target its promotions.
• BMW prospects appeared to have been receptive to this program,
encouraging BMW to roll it out on a larger scale.
Telling Them What They Want to Hear
The Weekly Standard is a conservative weekly political journal. The
magazine conducts extensive direct-mail campaigns using rented lists,
most of which are political fund-raising and subscription lists. For help
in increasing its subscriber base, it turned to Brian Brilliant at BMD in
Alexandria, Virginia.
The source of a list often demonstrates the mindset of the people on
that list. To leverage the limited data on potential customers, BMD’s
305
306
strategy was to insert a variable headline at the top of the letter that
would relate in a marketing sense to the list from which the recipient
came. For example, if John Doe’s name came from a military list, then
he would get a headline such as: “John Doe, we have some shocking news
about American’s decaying military.” Seven headlines were created to
match up with the rented lists. As you might expect, there were some
lists that didn’t fit any specific profile. They received a generic headline.
The headlines that were not selected for each prospect were then used
in the first paragraph as part of a variable message describing the variety
of subject matter that readers might find in the publication.
How well would such an approach work? To find out, Brian created a
control group for each list. The control group did not get a customized
headline, but instead got a generic headline about the magazine. Comparing the controls to the customized headlines, the custom letters had,
on average, a 24.8 percent better response rate and a 62.3 percent better
pay-up rate. The agency attributed most of the boost in initial response
rate to the “openability” of the package because of the personalized
message. And it attributed the lift in pay-up rate to the fact that subscribers were told up front that the magazine contained content that
would interest them. The method worked.
Cluster Coding
For the past 15 years, Claritas has done a brilliant job of dividing the
U.S. population into 62 clusters with catchy names like Shotguns &
Pickups and Pools & Patios. These clusters roughly reflect household
income and living style. For some products (although not for most), the
cluster someone lives in has something to do with the products that
person tends to buy. People in some clusters drink imported beer.
People in others drink only domestic. Folks in some clusters read the
New York Times. Folks in others read the New York Post. People in still
others do not take a newspaper at all. If your product can be shown to
be particularly attractive to people in certain clusters, you can use
this in your retention program. In a previous book, I described how
the Globe and Mail in Toronto made highly profitable use of cluster
coding to determine who read its newspaper and who didn’t. With that
knowledge, it was able to concentrate its telemarketing on the correct
postal walks, saving millions of dollars per year in acquisition costs.
Should you use cluster coding? First of all, remember that cluster
coding works only with consumer products. Second, consider whether
your product appeals particularly to a certain class of people. If you
think that it does, spend the money to have cluster coding affixed to
20,000 of your customers, ranked by revenue, and see if the clusters
show you anything worthwhile. Any good service bureau can have
cluster codes affixed to your file for about $120 per thousand, so your
20,000 will cost you a minimum of about $4000. In most cases, the
clusters won’t show you anything at all. If they do, you can use them to
make some money. Let’s assume that you have been sending direct mail
to about 1.2 million customers per month, for which you spend about
$650,000. Table 13-2 shows what your results might look like.
It is costing you about $30 to acquire a customer. This is your control. Now let’s try to beat that. We are going to select prospects for
mailing mainly from the correct clusters, with some others added to make
sure you are right. Your mailing next month could look like Table 13-3.
Table 13-2 Results before Using Cluster Selection
Nonselected
direct mail
Clusters with over
20% penetration
Clusters with 5% to
19% penetration
Clusters with less
than 5% penetration
Total customers
Cost
Sales
Rate
Cost per
sale
219,499
$111,944
7,463
3.4%
$15.00
452,991
$231,025
9,513
2.1%
$24.29
601,223
1,273,713
$306,624
$649,594
4,810
21,786
0.8%
$63.75
$29.82
Table 13-3 Results after Using Cluster Selection
Selected
direct mail
Clusters with over
20% penetration
Clusters with 5 to
19% penetration
Clusters with less
than 5% penetration
Total customers
Cluster coding
Total costs
Cost
Sales
Rate
Cost per
sale
950,000
$484,500
32,300
3.4%
$15.00
200,000
$102,000
4,200
2.1%
$24.29
50,000
1,200,000
1,200,000
1,200,000
$25,500
$612,000
$144,000
$756,000
400
36,900
0.8%
$63.75
$16.59
36,900
$20.49
307
308
You are still mailing 1.2 million pieces, and you are spending about
$100,000 more than before, as your costs include $144,000 for selecting
prospects by cluster. Even so, your cost per sale is down by $9, and you
have gained 15,000 more customers on the same 1.2 million mailing. This
shows the value of cluster coding, if it works for you.
Why did you do any mailing to clusters where your penetration was
less than 20 percent? That is a good question. Self-preservation is the
answer. As database marketers, we always have to justify what we do to
management, which questions everything. “Why did you spend this
$144,000?” your managers will ask. “Because I knew from previous tests
that cluster coding would work,” you answer. “Then why did you mail
to prospects in the lower-performing clusters?” they will ask. You reply,
“Because you would never have believed me if I hadn’t tested some of
these lower-performing clusters. Now you do.”
Prospect LTV
When you do acquisition, you want to know the LTV of the prospects
that you are trying to acquire. But how can you know anything about
them, since they have never bought anything from you?
Many companies have been quite successful at this kind of analysis.
First, you develop the LTV of your existing customers. Segment them
by demographics, and develop the LTV of each group. Next, you
append the same demographics to your list of prospects. Experian,
Donnelly, Acxiom, and other companies will be delighted to append
demographics (age, income, presence of children, home value, ethnicity, years at present address—some 200 different pieces of data) to your
prospect file. They will even find prospects for you that have the demographics that you are looking for. Then you make the assumption that
prospects that look just like your best customers will behave like your
best customers and will have a similar LTV.
Armed with this prospect file, your next step is to test your marketing skills. Can you develop the right approach through mass marketing
ads, letters, phone calls, or emails to win a significant number of these
prospects as customers? The final step will be to determine whether
Table 13-4 Selling to Segments
Customer
3-year
LTV
Group
Group
Group
Group
Total
A
B
C
D
$865
$522
$217
$ 14
Matching
prospects
mailed
Mailing
cost
Sales to
prospects
Prospect
response
rate
Cost
per
sale
400,211
200,011
50,034
50,055
700,311
$244,129
$122,007
$ 30,521
$ 30,534
$427,190
6,804
5,800
1,601
1,802
16,007
1.70%
2.90%
3.20%
3.60%
2.29%
$35.88
$21.03
$19.06
$16.94
$26.69
these prospect look-alikes really do act like the customers that they were
modeled after. When your tests are completed, do a direct-mail or email
marketing promotion. You can create a table for prospects that looks
like Table 13-4.
To create this table, you divided your customer base into four segments
based on lifetime value. You determined the demographics of these four
segments: income, age, presence of children, home value, and other
criteria. Using these criteria, you rented prospect names that matched
the profile of each segment, and you did a mailing to them. Your results
were interesting. The prospects that matched your best customers with
the highest lifetime value had the lowest response rate. Why should that
be? It is not at all unusual. Banks, for example, find that their best customers are the least responsive. Rich people get more mail than poor
people. They throw it away. In some cases, poor people are more likely
to respond. You have to decide whether you want responses (acquired
customers) or valuable long-term customers.
Group D is very interesting. It is costing you $17 to acquire customers who are worth $14. That does not make much sense. However,
that is what most companies are doing because they have not read this
book and they have not done the analysis. There is also another reason
why most companies are acquiring worthless customers: They are set
up for acquisition. They have a sales vice president who is compensated
on the basis of the number of customers acquired. He does not get
blamed if these customers are worthless. “It’s marketing’s job to make
money with these customers. I just find ’em and rope ’em in, and I do
a great job of it,” he says. Is this true in your company? Of course it is,
unless you have determined the lifetime value of your customers,
309
310
divided them into segments, and figured out which segments are winners and which are losers.
That’s what this book is all about—doing a thorough job of creating
a marketing database, and profiling it to find out important facts. Then
you use those facts to draw valuable and important conclusions.
Suppose Table 13-4 is a correct picture of the situation in your company. Your sales force is working hard, succeeding in bringing in a lot of
worthless customers, and getting bonuses for it. What can you do about
it? You have to do your homework, determine the lifetime value of the
various customer segments, and then put your findings on the desk of
top management. Convince management that the company may be acquiring the wrong people as customers. You will have a battle on your hands.
It may get rough, but if you have done the job right and you stick to your
guns, someone will recognize it. Then you have to work to change the
situation. After all, the purpose of an enterprise is to make a profit, not
just to acquire customers. We all know that, but we don’t act on it.
In a previous book, I described the situation I found at a client that
sold credit insurance for credit cards. A profiling exercise was conducted
that divided the U.S. population into seven sectors. Analysis showed that
only three of the seven sectors ever bought credit insurance. Armed
with this information, the company went to several banks (its customers)
and suggested to them that they stop telemarketing to these unresponsive segments, since the sales did not justify the expense. What did each
bank’s vice president of insurance respond? “Nuts. Call them all.” You
see, these vice presidents were compensated on the basis of the number
of customers acquired, not the cost per acquisition or the profits from
the insurance. So the insurance company went on calling everybody.
Selling Timeshares
Timeshares have become very popular in the past 10 years. More than
a million families own a week at some resort where they can go for a
vacation. RCI, a part of Cendant, makes a market in these timeshares;
owners can deposit their timeshare week with RCI and, in return,
receive a week at some other resort somewhere else in the world.
Because of RCI, some families rarely return to the timeshare that they
originally purchased.
The method whereby these timeshare owners are originally acquired
has become quite scientific and interesting. Most developers of timeshares sell tours as an acquisition method. A tour is a weekend vacation
that is offered to a family at below market rates—at prices as low as $50
for a 3-day family vacation at a hotel that is near a timeshare property.
While they are at the hotel, as part of their payment for the weekend
vacation, families are asked or required to take a tour of the timeshare
and hear a pitch on the value of owning a timeshare. Most tour operators
report that about 10 percent of the families that take a tour end up buying a timeshare, at prices ranging from $10,000 to $25,000 for 1 week.
The process of acquiring timeshare owners, therefore, has become
a process of selling tours. If you can get people to take a tour, you can
be confident that one in ten will make a purchase. The methods for tour
selling have become very scientific. Direct mail is used to offer households within 90 minutes’ driving distance of a timeshare resort a weekend
at a nearby hotel at a fixed below-market price. The response rate to
such direct-mail solicitations is about 4⁄10 of 1 percent. Telemarketing
then takes over to sign the respondents up for a tour. Here, again, scientific methods are used. Those who refuse a tour in June will be called
again about touring in July, or August, or some other month. “Call until
they tour or die” is the basic philosophy. And it works. Timeshares are
one of the most profitable industries in the United States.
Some interesting lessons have been learned in the course of acquiring
timeshare customers. These lessons can be useful in other industries.
For example:
• Vacation packages can be sold with a hook (a requirement that the
family listen to a pitch about the timeshare) or no hook (no pitch
requirement). Statistics show that no-hook recipients are more likely
to buy a timeshare than hook recipients. Why should this be so? A
hook family has steeled itself in advance against the timeshare pitch:
“OK, we will take their vacation, but we are certainly not going to
plunk down money for a timeshare.” The no-hook family’s thinking
is entirely different. Without the hook, the family members arrive
311
312
thinking, “Isn’t this great! A weekend in Orlando for only $69, and
no responsibilities other than to have fun.” While they are there,
they say to themselves, “Well, these nice people have given us this
great vacation. We might as well see what they have to offer. Maybe
we would like to own a place here in Orlando.”
• Tour direct mail that offers a specific date on the outside of the
envelope has a better response rate than offers that say, “Come to
Orlando this summer for a week on us,” with no specific date
mentioned. Why should this be so? If families think that if they
don’t accept this date, they will miss out on the vacation altogether,
they are probably more motivated to act immediately than those who
think, “Well, let’s see if we can find a week when it is convenient
for all of us to go together to Orlando.” After a couple of days,
they have forgotten the offer entirely. Direct mail must have some
urgency about it, or it doesn’t work well.
• Ugly one-color offers pull better than beautiful four-color offers.
Why? This is one of the great mysteries of direct mail. Sometimes
you don’t know the reason, you just know the results.
Nonprofit Fund Raising
How does a nonprofit fund-raising organization go about testing and
refining its methods? An outstanding example of this type of database
marketing is that of the Eastern Paralyzed Veterans Association (EPVA),
which has used a database to guide its mailing very successfully for many
years. Its database is managed by CSC Advanced Database Solutions.
To illustrate EPVA’s methods, the typical EPVA Christmas card
acquisition was aimed at about 2.4 million people selected from 72 different lists. Once donors had given the first time, they were sent an
appeal for a second purchase. This mailing was aimed at about 3.8 million
first-time donors. EPVA broke its database down into five groups:
ND
New donors, who made their first gift after the year-end
calculation
AD
L1
L2
L3
Active donors, who have made one or more gifts in the
previous 12 months
Lapsed donors who were not active in the last 12 months
prior to year-end
prior to year-end
prior to year-end
The EVPA also identified star donors, donors who at one time in
their history had given to three out of four consecutive card programs
and who had responded at least twice in the last 12-month period. This
is a subset of the active donor category.
The EPVA’s goals for the future were as follows:
1. Determine donor profiles. Who are our donors? What are the
attributes of a star donor? What are the differences between
multiple- and single-gift donors? What characteristic should we
be looking for when we rent prospect names?
2. Segmentation. What donors should be selected for what mailings?
Establish giving patterns and program interrelation. What modeling
approach should we pursue? Basically, we need to optimize our
mailing plans.
3. Lifetime value. Which acquisition lists pull the best over the long
haul? What effect do different test formats have on subsequent
mailings? Would donors produce more revenue over the long term
if they were mailed to less or more frequently?
4. Product analysis. Which products are the most effective in
retaining donors? How do reminder mailings affect donor
behavior? Are there groups of donors who consistently give
to only one or two products a year, and what would happen to
overall revenue if we mailed only those products to them?
5. Upgrading donors. What are the best methods of increasing
donors’ gifts?
313
314
Profile Plan
To achieve these goals, EPVA worked with CSC to undertake several
projects.
1. Donor profiles. Create a sample of 400,000 records, with 100,000
in each of the following categories:
• Star donors
• Nonstar donors—single donors
• Nonstar donors—multiple donors
• Nonresponding rental files
Use the demographic data available on these records to
determine the profile of these people, measuring their
• Gender
• Household income
• Age
• Housing type (own or rent and single family or multifamily)
• Ethnicity
• Presence of children
• Years at present address
• State
• Urban vs. suburban vs. rural
• DMA areas
• Occupation of head of household
• Education
• Marital status
• Number of adults in household
2. Create segments. EPVA planned to append Prizm codes to four
groups of 10,000 each:
• Star donors
• Nonstar donors—single donors
• Nonstar donors—multiple donors
• Nonresponding rental files
These data would help EPVA to determine which lifestyle groups
were more responsive to the EVPA message than other groups. The
results would be used for future list selection and donor mailings.
3. Lifetime value. EPVA decided to use the data in its database to
determine
• Which lists pull the best over the long haul. To do this, it
measured the response rate of each list used over the
previous 3 years, comparing the results to determine
whether the contributions were going up, going down, or
staying the same.
• The effect of frequency of mailing on the level of donation.
To do this, it set aside a test group of 20,000 two-time donors
and sent these donors only two mailings in the subsequent
year. An exactly similar group of donors received four
mailings. At the end of 1 year, EPVA could determine which
group had the highest contribution.
4. Product analysis. EPVA decided to compare the performance of
50,000 star donors and 50,000 nonstar active donors to determine
• What was the product that brought them to EPVA in the first
place?
• What products produced the best subsequent responses?
• How did each group perform by product?
Creating Referrals
Many companies have had success with referrals. The idea is to get your
existing customers to recommend that people that they are related to,
work with, or know also become customers. In other words, your customers become advocates. They tell their friends what a wonderful
company or product you have, and they get these friends to try your
product. Once someone has become a customer as a result of a referral,
something wonderful happens. These referrals who become new customers display, as a group, higher retention levels and higher spending
levels than the average customer who is recruited by other means. You
can prove this for yourself.
First, you have to set up a formal referral program. Tell your customers about it, and make it easy for them to generate referrals by
contacting a customer service rep or by going on the Web. Once you
315
316
have a new name, address, and email name, it is your job to get that
person to become a customer. However, you put the person’s ID
number into the referring customer’s database record, and you put the
referrer’s ID into the referred customer’s record. They are now tied
together. You can now measure their retention and spending rates and
their lifetime value.
Let’s say that you have done this, and you discover that your typical
customer has a 3-year LTV of $70, whereas your average referral has a
3-year LTV of $100. What can you do with that information? You can
use it to create a special referral incentive program. America Online has
done this. At various times, it has rewarded customers with $25 or $50
for recommending someone who also becomes a customer. This is a
wonderful example of successful database marketing. There are two
parts to the AOL story. In the first place, AOL has to spend a lot of
money to acquire customers. It places ads on TV, sends out CDs in the
mail, puts CDs in tens of thousands of stores in POS displays, and
recruits customers through partnership arrangements. This must be
expensive. I would imagine that it costs AOL at least $100 to acquire a
new customer. Once acquired, however, these customers pay AOL $23
per month forever. What is the marginal cost of servicing a new customer?
Not very much. Once you have the system and software in place, it is
not that expensive to add a new customer—perhaps $5 per customer per
month. At that rate, AOL is doing very well if it is spending $100 to
acquire customers who produce $18 in net revenue per month.
Of course, some of these new customers do not stay. They cancel
after a few months. This is true of all customers and all types of companies. That is where the referrals come in. Referrals are much less
likely to cancel than people who are recruited by other means. Why?
Because they know someone who is on AOL—the person who referred
them. They may have joined to send that person email. They will some
day have to face the question, “Well, how do you like AOL?” They don’t
want to have to say, “Well, I tried it, but I couldn’t get it to work” or “I
didn’t like it” or “It was a waste of my money.” They are not going to
want to say these things to their friends. So they are more likely to remain
as customers than people who have no such connections. For this reason,
AOL can afford to give away a little money to acquire such people, and
so can you.
Table 13-5 Regular-Customer LTV
Year 1
Year 2
Year 3
Customers
Retention rate
Spending rate
Revenue
100,000
60%
$275
$27,500,000
60,000
65%
$285
$17,100,000
39,000
70%
$295
$11,505,000
Cost rate
Costs
Acquisition/retention cost ($120/$20)
Total costs
60%
$16,500,000
$12,000,000
$28,500,000
55%
$ 9,405,000
$ 1,200,000
$10,605,000
54%
$6,212,700
$ 780,000
$6,992,700
Profit
Discount rate
NPV of profit
Lifetime value
($1,000,000)
1.00
($1,000,000)
($1,000,000)
($10)
$6,495,000
1.07
$6,058,769
$5,058,769
$50.59
$4,512,300
1.15
$3,926,525
$8,985,294
$89.85
Do your homework and find out how much you are spending to
acquire customers. Keep track of who is referring whom. Find out how
much more valuable referrals are than regular customers. Then set up a
program to share that profit and to reward your customers for referring
other people who become customers. Table 13-5 shows how regular
customers might look on a LTV chart.
Your regular customers cost you $120 to acquire and $20 to maintain
once they are acquired. Your retention rate begins at 60 percent. These
customers are worth $90 in the third year. Now let’s look at referred
customers (see Table 13-6). Let’s say you can generate 10,000 referrals.
As you can see, these referred customers have a much higher retention
rate. As a result, even after we reward our regular customers with $50
per referred customer, these customers have a third-year LTV of $206
instead of $90. They are really profitable customers. You will discover
the same thing. Figure it out, and design a referral program.
Summary
• From the Key Equipment case, we learned how to
• Create relevant and highly targeted prospecting lists that the
lead managers would believe to be credible.
317
318
Table 13-6 Referred-Customer LTV
Year 1
Year 2
Year 3
Customers
Retention rate
Spending rate
Revenue
10,000
80%
$275
$2,750,000
8,000
82%
$285
$2,280,000
6,560
85%
$295
$1,935,200
Cost rate
Costs
Referral fee/retention fee ($50/$20)
Total costs
60%
$1,650,000
$ 500,000
$2,150,000
55%
$1,254,000
$ 160,000
$1,414,000
54%
$1,045,008
$ 131,200
$1,176,208
$600,000
1.00
$600,000
$600,000
$60
$866,000
1.07
$807,836
$1,407,836
$140.78
$758,992
1.15
$660,462
$2,068,298
$206.83
Profit
Discount rate
NPV of profit
Cumulative of NPV profit
Lifetime value
•
•
•
•
• Develop direct-mail communications from each lead
manager that permit effective telemarketing follow-up.
• Have outbound telemarketing associates work closely with
the lead managers.
• Create a lead-generation database that tracks and identifies the
business generated by the lead managers from this program.
From the SalesLogix case study, we learned how to develop
• An online event registration system.
• A global campaign information database into which marketing
managers input campaign details through a Web interface.
From the BMW study, we learned how to
• Set up a 6-week communication program aimed at targeted
prospects.
• Use variables to score prospects.
• Deliver personalized content and information to each
responder.
From the Weekly Standard, we learned to modify messages to
prospects based on what we have learned about those prospects.
From the timeshare study, we learned a two-step process: Sell them
a tour, and let them buy the timeshare while they are on the tour.
So what have we learned overall from these studies?
• Long-term customer loyalty should be a prime consideration
during customer acquisition. It rarely is. You can test this in
your company by determining the LTV of your customers
by segment.
• Mass marketing may be the best method of acquiring some
customers.
What Works
• Profiling and modeling customers and looking for prospects that
match the profile of profitable customers.
• Providing communications from a person who will be contacting
the prospect.
• Providing personalized communications that reflect information
that we have about the prospect.
• Using telemarketers to support field sales personnel.
• Using a two-step process to zero in on those who are most likely
to buy.
• Giving prospects a direct and simple offer with a deadline. Too
many choices reduce response.
• Using the Web as a means for keeping everyone in the company
up to date on the acquisition process.
What Doesn’t Work
• Acquiring customers by giving discounts. It works in the short
term, but the customers do not stay with you.
• Assuming that every acquired customer is valuable. Many of them
can be unprofitable.
• Concentrating on acquisition to the exclusion of retention.
• Hanging on to customer names and not exchanging them with
others.
319
320
Quiz
1. In the prospect penetration case study, what can you use the
target potential index for?
a. Determining which customers are worthless
b. Determining your retention budget
c. Deciding which prospects you should aim at acquiring
d. Nothing, since it is a worthless idea
e. Scrap paper
2. What factor was not used to select prospects in the Key
Equipment case?
a. Major industry
b. Products purchased
c. Number of employees
d. Year company started
e. Sales volume
3. In the SalesLogix case, which of the following was not listed as
being a problem to begin with?
a. There were no e-marketing campaigns.
b. The company lacked high-technology products.
c. There was no lead-management system.
d. Telemarketing was not integrated.
e. The cost per lead was $500.
4. Which of the following was not one of the results of the
SalesLogix case?
a. Lead volume grew to 9600 per quarter.
b. Online events grew to 12 per month.
c. Telemarketing was outsourced to a better bureau.
d. The system filtered out 25 percent of raw leads.
e. There was 100 percent lead tracking.
5. In the BMW study, which one of the following was not done?
a. Using Experian data in regressions.
b. Delivering personalized content to prospects.
c. Assigning LTV to each customer and participant.
d. Using variables to score prospects.
e. None of the above; all these things were done.
6. List four product categories that cluster coding probably works
for and four that it probably doesn’t work for.
7. Which of the following is probably not true of most companies?
a. The company pays bonuses for customers, not for longterm customers.
b. The company does not know whether new customers will
be long-term or not.
c. The company has not computed the LTV of customers or
prospects.
d. The company is highly customer-centric.
e. Marketing and sales are going in different directions.
8. Why did the Weekly Standard custom approach work?
a. Customers are dumb.
b. Customers like to have their views agreed with.
c. It was a fluke. Most such copy fails.
d. The numbers are probably fake.
9. In the timeshare case, if no-hook tours perform better, why
does anyone sell hook tours?
a. The data are wrong. Hook tours are better.
b. Hook tours are traditional.
c. It takes time to change an established system.
d. It is hard to persuade people that what they are doing
is wrong.
e. Answers b, c, and d.
10. What are three things that msdbm did for BMW that helped to
solve the problem?
321
C
H A P T E R
14
Profile Marketing
F
or some time, creating customer profiles by appending data from
outside sources to a database of customers and prospects has been
a thriving business. The basis for this industry is the U.S. Bureau of
the Census, which gathers data from every household in America
every 10 years, the last time being in 2000. For most people, the census
collects only a minimum amount of data: name, address, age, gender,
and ethnicity. However, for 2 households out of 14, the census uses
the “long form,” which asks many questions about income, value of
home, type of plumbing, occupation, and so on—over 200 pieces of
data in all. These data are arranged for sale to businesses that are
interested. Before providing this information to the business community, the Census Bureau scrambles the data by averaging the answers
from the 2 households that filled out the long form and attributing
these data to all 14 households in the group. Because of this, you cannot
find out Arthur Hughes’s actual income or home value from the census
data. What you get appended to Arthur Hughes’s record is the average
of the data given by the 2 households out of 14 in the area in which
Arthur lives.
This isn’t a bad estimate because as America has evolved, most people
live in subdivisions in which the houses are all of similar value. As a
result, the incomes of the residents of each group of 14 houses are likely
to be similar, as are their ages, number of children, and so on. Only a
small percentage of Americans today live in diverse neighborhoods in
which poor and rich or old and young live side by side.
322
P r of i l e M a r k e t i n g
To the census information, some large data companies, such as
Equifax or Trans Union, add financial data derived from credit flows so
that they can estimate a household’s actual income and spending habits.
State driver’s license and vehicle registration information is used to
determine a person’s actual age and the type of car he or she drives.
Catalogers’ data files are used to determine responsiveness to direct-mail
offers. Most communities have developed new mover files—databases
of people who have just moved in. The U.S. Postal Service provides the
National Change of Address (NCOA) system, which corrects addresses
of households and businesses. By putting all this information together,
it is possible to create a pretty accurate profile of any household or individual if you know the household address.
Profiles are used to create customer and prospect marketing segments.
You can divide your database into such groups as
•
•
•
•
•
•
•
•
Teenagers
Young marrieds
Families with small children
Single women
Families with college-age children
Empty nesters
Seniors
College students
Once you have done this, you can check the national average for each
category against your own customer database. You can get a picture of the
type of person who buys your product and the type who does not. When
you know this, you can greatly reduce your marketing costs by trying to
reach only the most likely households instead of every household.
What can profiles tell you? Here are some sample profiling results
in which a company compared its customers to the national average.
To begin with, Figure 14-1 gives a profile by age range. It shows that
this company’s customers are clustered in the 35–54 age range, with
the percentage of customers in this age range being much higher than
the national average. Seniors are certainly not avid consumers of this
company’s products.
323
324
Figure 14-1
Profile by Age Range
company overall
national average
40%
35%
percent of total
30%
25%
20%
15%
10%
5%
0%
18–24
Figure 14-2
25–34
35–44
45–54
55–64
65+
Actual Age Range of Customers
60%
percent of file
50%
40%
30%
20%
10%
0%
Under 13
13–17
18–24
25–34
Over 35
These numbers were derived from appended data that reflect the age
of the head of the household. The company also asked its customers
what their actual age was. The results were quite different, as shown in
Figure 14-2. Here the company’s customers actually were quite
young—between 13 and 24. While Figure 14-1 gave the age of the head
of the household, most of the company’s customers were children in the
Figure 14-3
Income Ranges
company overall
national average
percent of total
30%
25%
20%
15%
10%
5%
0%
Un
de
1
r$
5,
00
$1
0
5,
0
1
–$
00
9,
99
$2
9
0,
0
2
–$
00
9,
99
$3
9
0,
0
3
–$
00
9,
99
$4
0,
9
0
4
–$
00
9,
99
$5
9
0,
0
7
–$
00
4,
99
$7
9
5,
0
9
–$
00
$
9,
99
0
10
,0
9
–
00
2
$1
4,
99
9
$
5
12
,0
00
+
household. Conclusion: If you want to know the real facts, ask your customers directly. Appended data do not tell the whole story.
What is the income range of your customers? Figure 14-3 gives the
results from the same company, using appended head of household data.
What can we learn from this graph? The consumers of this company’s
products live in households in which the income is close to the national
average. Poor people do not buy the products. Those with incomes over
$50,000 clearly do.
In terms of gender, the company’s products appeal more to women
than to men, as Figure 14-4 shows. Nationwide, women make up about
51 percent of the population. But 62 percent of the consumers of this
company’s products are women.
How Modeling Can Help
Modeling is complicated and expensive. It usually requires the appending
of external data, such as age, income, and presence of children. It requires
expensive software manipulated by skilled analysts. The minimum cost
for a model is usually $25,000, and some companies have spent well over
325
326
Figure 14-4
Profile by Gender
company overall
national average
70%
60%
percent of total
50%
40%
30%
20%
10%
0%
female
male
$200,000 for a complex model. So why does anyone do modeling?
Because, for large projects, it may pay off and give you results that are
better than those you could get without the model.
Summit Marketing Group in St. Louis did marketing for a major
property and casualty insurance company. The P&C insurance market
is very competitive. Hundreds of different companies offer automobile
insurance, and customers continually shop for the best deal. The pressure
to squeeze profits from every marketing dollar is intense. To help solve
the problem, Summit developed an optimization model that was
designed to maximize profits for each marketing campaign.
To acquire new automobile insurance customers, the P&C company
sent mailings to approximately 1.2 million households in a 10-state area
each month. The mailing list was drawn from a database of 9 million
households that included response and sales history, demographics, and
credit scores. To maximize profits, Summit focused on return on promotion (ROP). ROP is a discounted cashflow technique that is used to
measure the net present value of a marketing investment. It is theoretically similar to ROI. ROP considers response probabilities, conversion
rates, risk probabilities, acquisition costs, and other marketing decision
factors. The goal of the model is to determine which factors are most
important in creating the largest return on promotion, and to use these
factors in selecting the 1.2 million names to receive the mailing each
month. Since the company mailed every month, Summit had the previous months’ history to use as a base. The objective was to use the
model to get better and better at selecting the names to be mailed to,
in order to continually improve profits.
The modeling process produced some surprising results:
• The best candidates were not necessarily the best responders. The
best candidates probably already had insurance somewhere else and
were unwilling to shift.
• The best responders were often the least creditworthy, since they
have trouble getting insurance.
• Since the model mixes lifetime value, creditworthiness, and
probability of response, the top response deciles are not
necessarily the most profitable. The goal was profit, not response.
Table 14-1 shows the results of one month’s mailing, comparing the
previous month’s selection methods with those used in the current
month. The control group is what they were previously doing each
month. The optimized group is the new way of doing exactly the same
thing with better results.
Notice that the control group had higher revenue per sale ($1630)
than the optimized group did ($1360). The optimized group compensated for this by having a higher response rate and a higher sales rate.
Table 14-1 Return on Promotion
Total mailed
Cost of mailing
Number of responses
Response rate
Number of sales
Sales rate
Total revenue
Revenue per sale
Profit
Return on promotion
Control
group
Optimized
group
Percent
changed
Number
changed
1,264,571
$547,559
13,366
1.06%
1,599
12.0%
$2,605,603
$1,630
$95,896
18%
1,264,571
$547,559
16,090
1.27%
2,323
14.4%
$3,158,151
$1,360
$187,851
34%
0%
0%
20%
20%
45%
21%
21%
⫺17%
96%
96%
0
0
2724
0.22%
724
2.47%
$553,208
($270)
$91,955
16.8%
327
328
This allowed total revenue to increase by over $500,000 and profits to
increase by over $90,000 for one month’s mailing. This translates into
$6 million in added revenue and $1 million in added profit on an annual
basis. If the model cost $200,000 to run, it was still a worthwhile investment for the insurance company.
Getting Customers to Enter Their Own Profiles
The best data come from your customers themselves. Many of them
will be glad to give you information for free. In the past, sending survey
forms to customers by direct mail and receiving direct-mail responses
was a very expensive process. Only a small percentage of customers
(usually less than 7 percent) would respond to such a survey. When they
mailed in their survey forms, the data from these forms had to be
entered into a database by hand. The cost per completed survey was
substantial, as Table 14-2 shows. To justify this expenditure, you had
better have a very profitable marketing project.
All this has changed with the advent of the Internet. Today companies make survey forms an important part of their Web site. They
encourage people to come to the site and complete a survey form. The
costs are dramatically different. Assume that you already have a person’s
email address. You ask that person to complete a survey by email, directing him to your Web site. When he gets there, he does all the work.
Table 14-2 Cost per Survey
by Mail
Amounts
Requests sent
Cost per request
Cost
Response rate
Responses
Return postage
Data entry cost
Response cost
Total cost
Cost per survey
200,000
$0.60
$120,000
7.0%
14,000
$0.80
$0.35
$16,100
$136,100
$9.72
Table 14-3 Cost per Survey
by Email
Amounts
Requests sent
Cost per email
Cost
Response rate
Responses
Data entry
Data entry cost
Total cost
Cost per survey
200,000
$0.04
$8,000
7.0%
14,000
$0.00
$0
$8,000
$0.57
There is no data entry cost. A Web response customer profile will cost
you about $0.57, as shown in Table 14-3.
Why would customers want to give you this information? You have
to come up with some good reasons, or they won’t do it. One way to
get them to do it is to call these surveys preference profiles instead. The
benefit to the customer is that you will use these preference profiles to
provide customized service just for her. Figure 14-5 gives an example
of a preference profile.
Figure 14-6 shows how Macy’s asks for profiles. Why should you
register with Macy’s? Figure 14-7 tells you why.
While visitors to your Web site are filling out these surveys, you can
ask them their age, income, marital status, and so on. The information
you get will be far more accurate than appended data.
Figure 14-5
Disney Survey
329
330
Figure 14-6
Macy’s Survey
Effect on the Customer
Having customers fill out preferences profiles does more for you than
just provide information. It will change the behavior of your prospects
and customers. Once they have invested their time in filling out your
profile and have thought about the types of services they would like to
receive from you, they will think differently about your company. They
will be more responsive to your next offer, whether it is a direct-mail,
email, print, TV, or radio offer. You have moved them one step closer
to being profitable customers. You can prove that to your satisfaction
by comparing two groups of similar customers. To one group, you send
Figure 14-7
Reasons to Fill Out the Macy’s Survey
an email offer to complete their profile. To the other group, you do not.
Table 14-4 shows an example of the results.
If you look at this table, it is clear that those who completed your
profile turned out to be much better spenders than those who did not.
This does not necessarily prove that the profiles were worth it. Maybe
the people who filled out the profile were better customers to begin
with. But why did the nonresponders who were asked to fill out a profile
buy more than the people who were not solicited at all? Because you
communicated with them! Communicating with customers always
improves relations and sales—sometimes not as much as you would
hope, but always more than noncommunication. So if you want to determine the true result of your profiling operation, don’t just look at the
Table 14-4 Results of a Survey
Customers
Not completed
Annual sales
Sales per person
Completed
Annual sales
Sales per person
Test
Control
200,000
186,000
$4,836,000
$26
14,000
$1,512,000
$108
200,000
200,000
$3,200,000
$16
331
332
Figure 14-8
New York Times Survey
performance of the responders. Use a control group to see if the nonresponders also do better. They will. Figure 14-8 shows the kind of
information that the New York Times gets from its readers on the Web.
The Web Quiz
One of the most innovative and practical methods of customer profiling
through Web response is the Web quiz. People are encouraged to go to a
Web site to take a quiz. In most cases, the quiz concerns your product.
When they take the quiz, they get a score that tells them whether they
need your product and which of your products meets their needs.
Iomega, the maker of Zip storage disks, turned to Bob McKim of
msdbm of Los Angeles to design a Web quiz that would attract business
customers. Msdbm has perfected this technique. The company used direct
mail and email to prompt IT directors to come to Iomega’s Web site to
take a quiz (see Figure 14-9). The quiz asked the easy questions about data
storage given in the following list. As you look through these questions
and think about your answers, you will see that you are easily drawn into
the Iomega way of thinking. Each question has a yes or no answer.
1. At least once a quarter we make copies of all our data and store
them off site.
Figure 14-9
Web Quiz
2. My organization keeps at least three copies of our data and always
backs up to the oldest version.
3. My end users travel safely by carrying a disk copy of all their
important data.
4. My end users always carry a disk copy with them when they’re on
the road so that they can share large amounts of data.
5. My end users back up all their data at least once a week.
6. My end users all have a continuous backup product to protect
data between backups.
7. My end users employ a file version retention program to track
file changes.
8. My end users have actually practiced restoring data from backup
disks—just in case.
9. We use disk passwords to keep private data secure.
10. All our data disks are accurately labeled so we know what’s on
them and when they were created.
333
334
When the last yes or no has been checked, the user hits the Submit
button and sees the score (see Figure 14-10). Clicking on the Needs button brings up a further analysis of the user’s needs for Iomega products,
as shown in Figure 14-11. After a series of questions, another screen
provides a “Here’s What I’m Struggling With” button that brings the
user to a customized list of solutions that involve Iomega products.
Getting Users to Register
So far, the users are having fun on your Web site. You have their name and
their email name, which is wonderful because you can use them to follow
up. To get complete information on the users, however, Iomega wanted
their company data. It gave people an incentive to provide this by making
an offer. Bob McKim advised Iomega to offer respondents a drawing for
an Iomega HipDip Digital Audio Player. To enter the drawing, users had
to provide their address, their position in their company, information about
the number of PCs in the organization, and their authority over purchases.
Results
This kind of Web response can be very profitable. Overall, the direct
mail produced a response rate of 1.42 percent and a registration rate of
Figure 14-10
Results of the Quiz
Figure 14-11
The Fix for Your Problems
1.01 percent. These registrants provided their names and company
information and took the survey. The email promotion produced a
response rate of 3.49 percent with a registration rate of 1.95 percent.
It is significant that the email promotion was so much more productive. Figuring in the cost of the mailing, the results are shown in
Table 14-5.
Email promotion proved to be 11 times as cost-effective as direct
mail. Why would that be? There seem to be several reasons:
• The product was a computer product, so potential users all
had computers.
• The audience was IT directors, who were used to email and
were on the Web every day.
• The click-through process from an email takes only a second.
Trying to enter a URL from a direct-mail piece takes at
335
336
Table 14-5 Cost of a Survey
Unit cost
Cost for 50,000
Response rate
Number of responses
Registration rate
Number registered
Cost per lead
Direct mail
Email
$0.71
$35,500
1.42%
710
1.01%
505
$70.30
$0.12
$6,000
3.49%
1,745
1.95%
975
$6.15
least a minute—60 times as long. Compare these two
sentences:
To take the 1-minute Iomega Quiz and enter to win a prize,
click here.
To take the 1-minute Iomega Quiz, go on the Web and enter
www.msdbmdata/iomega/.
Which is easier?
What Works
• Appending data to your customer file to learn your customers’
demographics and create a profile
• Using the profile to find prospects that match the profile of your
best customers
• Asking customers to complete their own profile
• Using a customer’s profile to customize and personalize that
customer’s Web experience
• Using a quiz on the Web to find out about what customers want,
and then offering it to them
• Comparing your customer profiles to the national average
• Using a model to improve your response rate on promotions
What Doesn’t Work
• Getting people to fill out long, boring profile forms
• Assuming that the head of the household is the customer
Quiz
1. Go on the Web and find a company that asks you to complete
your personal profile. See what is good and bad about the way
this company goes about it.
2. Figure out how you could use customer profiles in your business.
How can you collect the data? How will you give your customers
an incentive to provide information?
3. In the P&C case study, what was the cost per piece of the mailing?
a. $0.33
b. $0.43
c. $0.53
d. $0.63
4. In the P&C case study, why didn’t the P&C company select those
with the highest revenue per sale?
a. This is a trick question: It did.
b. The optimized group had higher total sales.
c. The optimized group had higher response rates.
d. Both b and c.
5. Why are the best responders not necessarily the best people to
target in a mailing?
6. After a mailing to 100,000 customers, six of them telephoned to
say that they did not like the text of the mailing. On what basis
would you repeat the same mailing, despite these phone calls?
7. If you ask 400,000 people to complete a profile and only 2
percent respond by doing so, on what basis could you say that the
exercise was worthwhile?
8. Could you use a Web quiz in your business? Who would take the
quiz? What would be in it for them?
337
C
H A P T E R
15
A Farewell to the
Reader
T
hank you for making this journey with me. We have had some
laughs, and, I hope, we have learned something that will be useful
for our jobs. What have we learned?
Database Marketing Works
After two decades, database marketing is alive and well. It is being used by
thousands of companies throughout America, and to a certain extent
throughout the developed world. It is highly accountable and can be measured. You can prove that what you are doing is working, making customers
happy and making profits for your company. Here are some of the lessons
that have emerged from 14 chapters of theory and 40 case studies:
• Database marketing is not the same thing as CRM. You can call it
CRM, but that is rather like calling a terrorist a freedom
fighter. CRM is aimed at making the right offer to the right person
at the right time, based on massive amounts of data accumulated
in a warehouse. No warehouse can hold the amount of data on the
behavior of an individual that is needed if CRM is to deliver profits.
The warehouse would need to know what the customer is thinking,
what his values are, what he did yesterday, and what he wants to do
right now. CRM is a nice idea that does not work in practice and
338
A Fa r e w e l l to t h e R e a de r
wastes millions of dollars on an impossible quest. What does work
is database marketing, which is essentially aimed at divining the
behavior of customer segments and designing marketing programs
for those segments. This process is economically possible and can
be highly successful.
• Database marketing is incremental. Database marketing is a channel
for exchanging information with customers. For database marketing
to work, there have to be other channels that are delivering the
basic sales: mass marketing, retail stores, catalogs, interactive Web
sites, salespeople, distributors, agents, outbound telemarketing, or
direct mail. What database marketing does is make incremental
improvements in customer retention rates, response rates, referrals,
cross sales, frequency of sales, and average order sizes. Database
marketing builds relationships with customers that bind them to your
company and your brand. Done right, it makes customers happy
and gives them a feeling of identification with you, your products,
your company, and your employees. It helps prevent defections.
• The Internet has enabled database marketing to deliver on its promises.
The idea was always to use the database to communicate with
customers and make them happy. But direct-mail and telephone
communications are so expensive, in relation to the good that they
do, that they can be used only sparingly. Email to customers is
almost free. It is instant. It can be highly personalized, delivering
up-to-the-minute information. It is the best thing to hit database
marketing in the last 20 years.
• The Web is not a selling medium; it is an ordering medium. An entire
industry based on Web advertising and sales was created, wasted
hundreds of billions of dollars, and then died within 5 years. What
did we learn from this? The Web is a passive medium, not an active
one like TV, radio, telemarketing, direct mail, or retail stores. It is
not a very powerful advertising medium. People will buy products
and services over the Web, but they first have to be stimulated by
something else: catalogs, direct mail, or mass marketing. As an
ordering medium, the Web can be superb. It cuts the cost of a
customer contact by more than $3. It can be tied to a database so
339
340
•
•
•
•
that each customer sees something different, customized for her,
when she clicks on a site. But she is unlikely to click on that site
without some outside stimulus.
People like communications. The heart of database marketing
is not the database but the customized and personalized
communications with customers that are created using the
database. Communications work. They keep customers from
defecting. They make people happy (and angry as well). They
help to sell goods and to increase loyalty, referrals, cross sales,
order frequency, and order size. Without an active and creative
communications program, the database is worthless.
Everything in database marketing can and should be measured.
Whenever you design a new strategy, program, or communication,
you must set aside a control group that does not get the Gold card,
communication, newsletter, offer, discount, premium, or special
service that is being extended to your test group. This is the
only way for you to know and prove to management that your
expenditures for database marketing are producing a positive
return on investment. If you are not constantly testing new ideas,
setting aside control groups, and measuring, you are not doing
database marketing.
Lifetime value is the key measurement technique for database marketing.
You must learn how to compute customer lifetime value and use
it in evaluating and designing your marketing strategy. LTV is a
forward-looking concept that includes spending rate, discount rate,
retention rate, referral rate, product costs, and marketing
expenditures. It is the only way to know how much you should
spend on acquisition and retention. It is an essential step to use
in getting your marketing budget approved each year.
Database marketing is not about discounts. Discounts are used in
mass marketing campaigns, and they work. They don’t work in
database marketing. Why not? Because the basic idea in database
marketing is to build a close personal relationship with each
customer that is based on quality, service, friendship, loyalty, and
communications. You would not give a neighbor $5 for helping you
carry a chest of drawers up your stairs. It would be an insult. Instead,
you offer your neighbor a cup of coffee or a beer and 15 minutes of
chat around the kitchen table. That is the kind of relationship that
database marketing creates. Discounts send the wrong message: We
are cheap guys whose basic product is overpriced. We want to buy
your loyalty. We don’t care about you. We care about your money.
• Customers should not all be treated alike. Some of your customers
are Gold. They deliver huge amounts of sales and profits. Many
customers of many companies are actually unprofitable. Why
should you work to retain the loyalty and sales of an unprofitable
customer? What should you do? First, create customer segments
and learn who your Gold customers are. Create special programs
and services just for them in order to retain them. Study the
segment just below Gold. Create programs that will get them to
move up. Study your bottom segment—the worthless people.
Figure out how to either reprice the services you are giving them
so that they are profitable or get rid of them altogether. If you
treat all customers alike, your service budget will be stretched so
thin that you will be unable to use it to modify customer behavior.
• Most successful database marketers outsource the construction of their
database. Why? Because it is cheaper and faster, and the product is
better. To take a ridiculous example, it would be possible to go to
auto parts suppliers and assemble a company truck from spare parts.
Since no one in your company has ever done this before, it would
take a year or more and be quite expensive, and the result certainly
would not perform as well as a production model bought from GM,
Ford, or Chrysler. But it would have the advantage that, as a result,
your staff would now know how to build a truck from spare parts.
The problem with this approach is that most companies are not in
the truck manufacturing business. They are in some other line of
work. Building a truck would be a costly diversion from their core
business. And, during that year, the company could not use the
truck to make deliveries.
The same principle applies to building a marketing database from
spare parts. No one in most companies (including the IT department)
has done this before. It takes a long time, since there is a learning
curve associated with each process. It takes time to study the
341
342
available software packages (there are scores of them) and to install
and learn to use the ones chosen. When the database is finished,
it will not work as well as one that was built by an experienced
service bureau that has already built dozens of them and has a
trained technical staff whose only job is building and maintaining
marketing databases. Finally, you cannot get any benefit from the
home-grown database during the year or more that is needed to
build it. You get the benefit only after it is up and running.
Once your database is up and running, it can be migrated to inhouse maintenance at any time. In fact, however, if the database is
built correctly, your marketers will be so happy with the results of
using it that they will forget about wanting to maintain it. When
you are driving a new BMW down the interstate at 75 mph, who
wants to stop and become a garage mechanic?
• Caller ID and cookies have become essential database marketing tools.
Caller ID is used by customer service reps to recognize customers,
call them by name, and see their history with the company on the
screen before they even answer the phone. This type of recognition
builds relationships and loyalty. Cookies are used the same way on
the Web. Web sites are configured differently for each repeat
visitor, based on that visitor’s expressed preferences on the previous
visit. Cookies build relationships and loyalty.
• Many customers will gladly give you their profiles over the Web. What
is a nuisance on paper or over the phone becomes a fun exercise
over the Web. The marketer can use the profiles to learn more
about customers and provide them with what they want (as opposed
to what the company wants to sell them). Profiles open up an
entirely new and productive channel for customer contact.
• Marketing databases today are stored in a relational format on a server
accessed by marketers over the Web. Many different marketers can
access the company customer database simultaneously, using IDs
and passwords so that they see only the data that they need. Because
of the Web, the database can be used by remote offices, salespeople
on the road, and even overseas branches of the company. Another
advantage of the Web as an access tool for marketers is that all the
marketer needs to have is a browser. She does not need to have any
database software on her PC.
• Becoming customer-centric is seldom an achievable goal. For the last
decade, marketers have talked about shifting their company from
being product- and brand-centric to being customer-centric. “Don’t
sell products. Find out what customers want to buy and sell them
that.” This is a wonderful idea, but few companies are doing it. If
we were really to become customer-centric, we would segment
our customers and create a manager for each segment. We would
give these managers greater authority than that given to brand
or product managers. In most cases, this does not work. The
compensation system within a company is very difficult to arrange.
If the Silver Segment manager does a good job, many of his
customers will become Gold, and he will lose them. How do you
give him an incentive to do that? Before you mouth platitudes
about becoming customer-centric, think it through. See if you
can make it work on paper. It won’t. Forget it, and do a good job
of database marketing within your current organization.
Summary
Database marketing is a method of getting incremental increases in
retention and sales when sales are conducted primarily through other
channels, but it is also much more than that. It is a way of making
people happy. People like to buy products and services. They like to
establish personal relationships with their suppliers. They like to be
recognized, thanked, and chatted with. They like to receive wanted
communications. You can do all this with database marketing using the
Web, phone, and direct mail. You cannot do these things with mass
marketing. Merchants used to recognize customers and communicate
directly with them in small retail stores, but most of the companies
using database marketing today are not proprietors of small retail stores.
What successful database marketing does is to bring back the close personal relationships that used to exist between customers and local stores
and that have been lost as our country and economy have grown. In
doing database marketing well today, you are making the world a happier,
more satisfying, and better place to live. And you will be making an
honest living doing it.
343
A
P P E N D I X
How to Keep Up with
Database Marketing
I
n the months ahead, you will need to stay abreast of the new
developments in our industry. Here are some ways of doing this.
Attend Conferences
National Center for Database Marketing, two conferences per year,
www.the-dma.org.
Direct Marketing to Business, www.dbmshow.com.
Other DMA conferences: Net marketing, brand, and others,
www.the-dma.org.
Read Magazines
DM News, www.dmnews.com.
Target Marketing, www2.targetonline.com.
Direct magazine, www.primediabusiness.com.
iMarketing magazine, www.imarketingnews.com.
Visit Web Sites
www.dbmarketing.com
344
H o w t o K e e p U p W i t h D ata b a s e M a r k e t i n g
www.computerstrategy.com
www.1to1.com
www.gartner.com
Read Books
The Complete Database Marketer, by Arthur Middleton Hughes (New
York: McGraw-Hill, 1996).
Strategic Database Marketing, by Arthur Middleton Hughes (New
York: McGraw-Hill, 2000).
The Loyalty Effect, by Frederick Reichheld (Boston: Harvard Business
Press, 2001).
Loyalty Marketing, the Second Act, by Brian Woolf (Greenville, SC:
Teal Books, 2002).
Customer Specific Marketing, by Brian Woolf (Greenville, SC: Teal
Books, 1998).
Data Mining Cookbook, by Olivia Rud (New York: Wiley, 2000).
Exchange Information
I am going to make a request of you as a reader. If you are using any of
the techniques in this book in your work and you have a wonderful
example of success, could you contact me at dbmarkets@aol.com and
let me know about it? I would like to publish your success in an article
and perhaps a later book. When we share our experiences in this way,
database marketing becomes more and more successful as a technique,
and I get to write more books.
Get Help
Many companies will help you to build your database or send emails.
One that I can highly recommend that you consult is www.computerstrategy.com.
345
Answers to Quizzes
Chapter 1
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
(c) Treating all customers alike
(a) Customer communications
(e) 0.4 percent
(d ) Marketing strategy
(a) Those customers in priority A and priority B
(b) The resources available to devote to the best customers will
be too small.
(d ) On the computer holding the database
(c) It made innovative use of the Web.
(d ) Those who do not get any of the test communications
(e) Not a good way to recruit new customers who last
Chapter 2
1. (e) Increased employee loyalty is essential to increased customer
retention.
2. (c) The marketing costs
3. (a) CRM software is not available today.
4. This paragraph cannot be converted into English. It is
incomprehensible.
346
A n s w e rs to Q u i z z e s
5.
6.
7.
8.
9.
10.
(d ) 50 percent
(b) Build a data warehouse
(b) Net present value
(c) The basic product sales are produced by other channels.
(d ) Because the compensation system cannot be changed
(c) Mass marketing
Chapter 3
1. (b) The public feared that the government was going to go after
the Web next.
2. (c) The Web is a passive medium.
3. (b) Because of the economics of the supermarket industry
4. (a) There are none. It is a wonderful idea.
5. (b) The e-Citi customers could not use Citibank branches.
6. (c) Many competitors copied its ideas and stole its
customers.
7. (c) Keep customer communications to a minimum.
8. (b) Don’t try to compete with telephone customer service.
9. (e) Customer service
10. (e) Consumer product sales
Chapter 4
1. D ⫽ (1 ⫹ i * rf )n ⫽ (1 ⫹ 0.06 * 1.4)2 ⫽ 1.175
2. $45,447,221/1.175 ⫽ $38,678,486
3.
Customers
Retention rate
Year 1
Year 2
Year 3
312,886
219,445
188,994
4. (a) Make it go up
5. (d ) Both a and b
6. (b) Movement in the lifetime value in future years
347
348
7. (d ) The one with the highest value to the company, which is
determined by multiplying the probability of purchase and the
profitability if purchased
8. (a) Revise your marketing program.
9. (a) Work to retain segment B and increase the LTV of segment A.
Chapter 5
1.
Year 1
Year 2
Year 3
Customers
Retention rate
Spending rate
Revenue
400,000
62%
$320
$128,000,000
248,000
70%
$340
$84,320,000
173,600
74%
$350
$60,760,000
Cost rate
Costs
Total costs
65%
$ 83,200,000
$ 56,000,000
$139,200,000
63%
$53,121,600
62%
$37,671,200
$53,121,600
$37,671,200
⫺$11,200,000
1
⫺$11,200,000
⫺$11,200,000
⫺$28
$31,198,400
1.22
$25,572,459
$14,372,459
$35.93
$23,088,800
1.35
$17,102,815
$31,475,274
$78.69
Profit
Discount rate
NPV of profit
Lifetime value
2.
Year 1
Year 2
Year 3
Customers
Retention rate
Spending rate
Revenue
400,000
66%
$332
$132,800,000
264,000
74%
$352
$92,928,000
195,360
78%
$362
$70,720,320
Cost percent
Costs
Marketing cost ($6)
Total costs
65%
$ 86,320,000
$ 56,000,000
$ 2,400,000
$144,720,000
63%
$58,544,640
62%
$43,846,598
$ 1,584,000
$60,128,640
$ 1,172,160
$45,018,758
⫺$11,920,000
1
⫺$11,920,000
⫺$11,920,000
⫺$30
$32,799,360
1.22
$26,884,721
$14,964,721
$37.41
$25,701,562
1.35
$19,038,194
$34,002,915
$85.01
Profit
Discount rate
NPV of profit
Lifetime value
3.
Old LTV
New LTV
Difference
400,000 customers
Year 1
Year 2
Year 3
−$28.00
−$29.80
−$1.80
−$720,000
$35.93
$37.41
$1.48
$592,262
$78.69
$85.01
$6.32
$2,527,641
4.
Mailing
Response
Response revenue
Less mailing
Per name
Offer price
Sales
Commission
100,000
1,200
4
$0.05
1.20%
$78.69
$5,000
$5,000.00
$94,425.82
$89,425.82
$0.89
$0.45
$1.79
25%
$1.34
5. Answers will vary. You have to do the work.
6. Any five of the six following answers are correct:
a. Low margin.
b. Coupon redemption is very slow.
c. To make a profit, you need repeat sales (low margin). You
can’t redeem a coupon for every sale, or you would go broke.
Therefore, you never find out about the subsequent sales.
d. There is no way to figure out if you are succeeding.
e. Discounts don’t produce loyalty.
f. Manufacturers won’t buy the names.
7. Answers will vary.
Chapter 6
1. (c) Cooperation between IT and Marketing
2.
Mailed
100,000
Response
1160
Cost
Total
$0.05
Response rate
$5000.00
Cost for each
1.16%
$4.31
349
350
3. (b) Mutual fund offer. It is easier to get people to do more of
what they are doing than it is to get them to do something
else.
4. (d ) Emails are so inexpensive you can concentrate on
maximizing profits.
6.
Telemarketing
Email
Cost
each
Contacts
$5.00
$0.06
100
100
Cost
Order
rate
Cost
per
sale
Sales
ROI
$500.00
$6.00
8
3
$62.50
$2.00
$1600
$600
25.6%
300%
7. Name, email name, permission to use the email name, zip
code. You should also try to get at least one other fact, such as
customer preference for contact, income, age, or family
composition.
8. (d ) The control group
9. Sales ⫽ $150,000; ROI ⫽ 8.32; cost ⫽ $18,750
10. (d ) The subject line
Chapter 7
1. (a) People are curious about the offering.
2. (a) A high-tech flash page
3.
Mails
Click-through
Registrations
4.
5.
6.
7.
8.
9.
10.
Rate
Number
Dollars
$0.11
0.67%
48%
100,000
670
322
$11,000
$16.42
$34.20
(b) You can accurately predict response rates.
(e) None of the above; it proved all of them.
(e) To reduce churn
(e) Video clips
(b) Fax
(c) Email
(a) Phone
Chapter 8
1.
2.
3.
4.
5.
6.
(d ) Segments cost less but produce similar results.
(b) Preferences
(b) Credit card matching
(a) Web customer response
(c) Ignoring them
(b) Cross shopping was reduced by 35 percent (it increased by
35 percent).
7.
Teens
Control
20s and 30s
Control
40s and 50s
Control
Seniors
Control
Total
Number
promoted
Sales
rate
100,000
20,000
100,000
20,000
90,000
20,000
60,000
20,000
430,000
8.80%
7.00%
6.60%
5.80%
7.20%
6.00%
5.10%
4.30%
Lift
Lift in
sales
1.80%
1,800
0.80%
800
1.20%
1,080
0.80%
480
4,160
8. (e) 6
9. (a) Something that you can relate to and design programs for
10. (d ) After each promotion.
Chapter 9
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
(b) Existing customers were modeled.
(c) Prizm cluster codes
(c) SIC analysis
(e) 50 percent
(a) Vendors do the selling for their customers.
(d ) A rate calculated using the agent’s previous sales
(a) The Web
(e) All of the above
(a) The supplier
(c) Business customers usually pay long after shipment.
351
352
Chapter 10
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
(b) Clean up 80 percent of your files first.
(d ) Are not in priority C
(c) Birthday cards
(e) Geographic region
(a) The cross-sell model
(e) Brand managers fought over the more profitable customers.
(b) Customers were ranked by RFM with a maximum score of 125.
(a) The bank added 2000 new total customers each month.
(e) All of the above
(a) Reports produced in a few weeks
Chapter 11
1.
2.
3.
4.
(b) Customers like to participate in the selling process.
(e) All of the above.
(d ) Eliminate toll-free calls so that people must use the Web.
(e) None of the above; they are all rules for successful Web
customer service.
5. (a) Vendor-managed inventory
6. (b) The parts are warehoused in customers’ warehouses.
7. (e) More CSRs are needed.
8.
Year 2
Year 3
Customers
Retention rate
Basket size
Visits per year
Revenue
Year 1
200,000
45%
$60
2.10
$25,200,000
90,000
55%
$65
2.50
$14,625,000
49,500
65%
$70
2.80
$9,702,000
Cost percent
Costs
Total costs
65%
$16,380,000
$ 9,200,000
$25,580,000
62%
$9,067,500
60%
$5,821,200
$9,067,500
$5,821,200
⫺$380,000
1
⫺$380,000
⫺$380,000
⫺$1.90
$5,557,500
1.25
$4,446,000
$4,066,000
$20.33
$3,880,800
1.4
$2,772,000
$6,838,000
$34.19
Profit
Discount rate
NPV of profit
Lifetime value
9.
Year 2
Year 3
Customers
Retention rate
Basket size
Visits per year
Revenue
Year 1
200,000
49%
$64
2.50
$32,000,000
98,000
59%
$69
2.90
$19,609,800
57,820
69%
$74
3.20
$13,691,776
Cost rate
Costs
Retention cost ($8)
Total costs
65%
$ 20,800,000
$ 9,200,000
$ 1,600,000
$31,600,000
62%
$12,158,076
60%
$8,215,066
$ 784,000
$12,942,076
$ 462,560
$8,677,626
$400,000
1
$400,000
$400,000
$2.00
$6,667,724
1.25
$5,334,179
$5,734,179
$28.67
$5,014,150
1.4
$3,581,536
$9,315,715
$46.58
Profit
Discount rate
NPV of profit
Lifetime value
10.
Old LTV
New LTV
Difference
200,000 customers
Year 1
Year 2
Year 3
−$1.90
$2.00
$3.90
$780,000
$20.33
$28.67
$8.34
$1,668,179
$34.19
$46.58
$12.39
$2,477,715
Chapter 12
1. (a) Personalized catalogs
2. (a) It provides no sales lift.
3. To reach high end customers; to reduce ordering costs; to open
up another channel
4. They spent 33 percent more than regular customers.
5. Cross sales were up 100 percent.
6. They need to add:
a. Customer profiles capturing emails and customer
preferences and demographics. This reaches 20 percent
on the first round and an additional 20 percent on each
additional round.
b. One-click ordering for all customers completing their
profile, plus personalized emails and Web site opening
pages.
353
354
7.
8.
9.
10.
c. Email thank you for all orders, emails when the order is
shipped, email asking people if the order was satisfactory.
d. Personalized emails announcing the arrival of each
catalog, with click here for items that it is assumed the
customer is interested in.
e. Collaborative filtering to make suggestions on next best
product, leading to a cross-sale rate of 40 percent.
f. Advanced search feature that permits customers to find
products faster.
g. Live shopper support for people who want to have a
text chat with a customer service rep while they are on
the site, or who want to call on another line to talk to
a live agent.
It provided a window showing the address, phone number, and
directions to the nearest Macy’s store.
(e) None of the above; they were 80 percent higher.
(e) None of the above; catalog distribution increased by 20
percent per year.
(b) 26 percent
Chapter 13
1.
2.
3.
4.
5.
6.
(c) Deciding which prospects you should aim at acquiring
(b) Products purchased
(b) The company lacked high-technology products.
(c) Telemarketing was outsourced to a better bureau.
(e) None of the above; all these things were done.
Works: newspapers, mall location, luxury car sales, financial
services
Doesn’t work: business-to-business, nonprofit, packaged goods,
nonluxury cars
7. (c) The company has computed the LTV of customers or
prospects.
8. (b) Customers like to have their views agreed with.
9. (e) Answers b, c, and d.
10. It:
a. Performed regression and correlation analysis based on
customer-supplied and Experian data on current BMW
owners to identify unique independent variables.
b. Developed score ranges for variables based on frequency
distribution, standard deviation, and historic experience.
c. Scored samples of prospects and compared results.
d. Developed lifetime value tables and assigned a value to
each customer and participant.
e. Delivered personalized content and information to each
responder, whether that responder was a previous
customer or a prospect.
f. Used variables to score prospects; these variables included
purchase time frame, make of current vehicle, age,
income-to-price ratio, year of current car, source of
prospect lead, and debt-to-income ratio.
Chapter 14
1.
2.
3.
4.
5.
6.
Answers will vary.
Answers will vary.
(b) $547,559/1,264,571 ⫽ $0.43
(d ) Both b and c.
Because they may be the least creditworthy
If the mailing was profitable, you should not have your marketing
program derailed by six malcontents out of 100,000.
7. If the revenue from the information and sales to the 2 percent is
profitable after deducting the cost of the profile, the response
rate is unimportant.
355
Index
Acquisition, customer, 293–296
Acrobat, 148
Acxiom, 231–232, 235, 308
Advertisers, keeping, 111–115
Advertising:
selection of media for, 234
Web, 43
Affiliates, 278
AgLine, 116
Air Miles for Business, 116, 117
Amazon.com, 3, 39, 49, 51, 54, 87, 118,
120, 124, 127, 254, 256, 259–260,
275, 280, 281
America Online (AOL), 39, 40, 43, 134,
303, 316
American Airlines, xiii, 118, 221–223
American Express, xiii, 166
Andrew, Skip, xiii
Andrews, Scott, 242
Angara Reporter, 48
AOL (see America Online)
Appleseed, 282
Ashford.com, 278
At Home Corp., 48
Attrition rate, 11
Automatic trigger marketing, 227–231
Automobile industry, lifetime value in,
64–75
Avis, 56
Brostoff, Mike, xix
Brostoff, Scott, 188
Budget, 56
Business customers, 187–211
BenefitMall example, 195–200
inventory-management example,
208–209
next best product of, 190–193
Postcard Direct prospecting example,
200–203
pros and cons of selling to, 187
shipping partnership example, 209–210
and supply chain management,
204–208
top ten prospects, identifying,
188–190, 193–195
travel agent example, 203–204
Business-to-business catalogs, 285–290
customer profiles for, 286–288
email marketing for, 288–290
setup costs, 288
Business-to-business communications,
xiv, 109–110
Business-to-business marketers, 5
Business-to-business relationship
building, 109–110
Business-to-business Web transactions,
52–53, 256–258
Buy.com, 120
Bain & Company, 31
Bank of America, 280
Banking, 25, 50–51
Banner ads, 43
Barnes & Noble, 3, 39
Behavior, customer:
and customer surveys, 330–332
and loyalty programs, 224–225
Behavior changes, communications and,
116–118
Behavioral data, 166
Behavioral segmentation model, 232
BenefitMall.com, 195–200, 210, 212
Blockbuster Video, 240
BMD, 305
BMW, 302–305, 318, 320, 321, 342, 355
Boeing, 86, 87, 261, 263–266, 268
Borders, 39
"Bricks and click" merchants, 39
Brilliant, Brian, 305, 306
Brooks, 48–49
Call center software, 271
Caller ID, 3, 271, 342
Canadian Direct Marketing News, 164
Carlson Marketing Group, 155, 156,
157, 203, 225
Catalogs, 270–291
business-to-business, 285–290
and collaborative filtering, 280–282
and database marketing, 271–272
and lifetime value, 282–285
personalized, 272
success of, 270–271, 273
and the Web, 272–282
CDNow, 87, 120
Cellular service example, 239–241
Cendant, 310
Chrysler, 341
Chung, Sungmi, 33
Churn, 11–12, 239–241
Citibank, 61, 222, 347
Citigroup, 50–51
357
358
Inde x
Claritas, 13, 14, 306
Click-through rates, 280
Clough, Beth, xix, 120, 121, 123
Club membership example, 241–245
Cluster coding systems, 13, 294, 306–308
Coding systems, 13, 143–144
Collaborative filtering, 280–282
Color(s):
in direct mail, 312
Web, 282
Communications, 5, 6, 109–139, 340
and automatic trigger marketing, 229
and behavior changes, 116–118
and business-to-business relationship
building, 109–110
and consumer relationship building,
110–111
cost of (CRM), 27–29
customer-focused, 236
by email, 118–129, 132–138
and keeping advertisers, 111–115
relationship-building with, 197
and RFM, 115–117
StrideRite example, 130–132
and Web catalogs, 274
Compaq, 42, 187–188, 205, 210
Complete Database Marketer, The (Arthur
Hughes), 1, 345
Computer Associates, 134
Consumer relationship building, 110–111
Consumer Reports, 44
Control groups, 15–16, 252, 276, 278
Cookies, 7, 54, 276, 342
Coolbaugh, Karen, 242
Costs:
for B2B catalog setup, 288
of building data warehouses, 216–218
of CRM, 21, 26–29
of customer service, 8, 253–255
of vendor-managed inventory, 206, 207
of Web sites, 261–263
Country Inns & Suites, 204
Credit cards, 222, 270–271
CRM (see Customer relationship
marketing)
CRM Forum, 32–33
Cross-sell model, 232–234
Crow, Sheryl, 56
CSC, xix, 132, 188–190, 210, 211, 242,
245, 285, 312, 314
CSI (Customer Service Index), 158
CSRs (see Customer service reps)
C2It, 51
Customer acquisition, 293–296
Customer Connection, The, 109, 111
Customer involvement, 249–250,
258–259, 266
Customer management, 214–246
via automatic trigger marketing,
227–231
Customer management (Continued)
cellular service example, 239–241
club membership example, 241–245
and 80 percent rule, 215–218
via loyalty programs, 220–227,
231–235
need for restraint in, 215–216
risk/revenue analysis for cutting costs
of, 218–220
segmenting in, 235–239
Customer profiles (see Profiles)
Customer referrals, 315–317
Customer relationship marketing
(CRM), xiv, 16, 21–35, 168
assumptions underlying, 22
costs of, 21, 26–29
customer-centricity of, 25–26
database marketing vs., 23–24,
338–339
and decision to buy, 23, 24
failure of, 32–33
goals of, 22, 32–33
and IT, 22
and loyalty effect, 29–32
return on investment from, 26
and timelines/relevance of offers,
24–25
Customer response, 142, 149–155
Customer segments (see Segments,
customer)
Customer service:
costs of, 8, 253
and customer segments, 251–252
via the Internet, 253–267
live online, 276–277
and marketing, 250–251
Web setup of, 259–263
Customer Service Index (CSI), 158
Customer service reps (CSRs), 55–56,
59, 251, 252, 254–256, 281
Customer Specific Marketing (Brian
Woolf ), 227, 293, 345
Customer-generated profiles, 328–336
Customer(s):
distance between suppliers and,
249–250
focus on, 25–26
Dark Sun (Richard Rhodes), 281
Data, demographic vs. behavioral, 166
Data marts, 217
Data Mining Cookbook (Olivia Rud), 345
Data warehouses, 15, 216–218
Database cleanup, 215–216
Database marketing, xiv–xviii, 1–2
and benefit to customer, 2–3
BenefitMall example, 195–200
and catalogs, 271–272
changing approaches to, 4–6
Inde x
Database marketing (Continued)
by computer companies, 187–188
CRM vs., 23–24, 338–339
next best product identified with,
190–193
top ten prospects identified with,
188–190, 193–195
Database Marketing Institute, 106
Databases:
failure of, 15–17
types of, 13–15
D&B (see Dun & Bradstreet)
Dbmarketing.com, 106
Dealers, finding, 49–50
Decision support system, 240–241
Dell Computer, 7, 188, 205, 210, 254,
256–258, 268, 280
Deloitte & Touche, 80
Delta Airlines, 221
Demographic data, 166
Demographic information, 275
Dialog, 229
Digital Impact, 96
Direct Marketing Association, 119
Direct Marketing to Business, 344
Direct-mail response, Web response vs.,
149–152
Direct-marketing promotions, 5
DirectTR@K, 197
Discount rate, 67, 72–73
Discounts, 47, 293–294, 340–341
Disney, 329
DM News, 344
Donnelly, 14, 308
DSW Shoe Warehouse, 225
Dun & Bradstreet (D&B), 5, 13,
188–189, 191, 194, 195, 200,
210–212, 227, 286, 295, 296, 298
Dun's Major Industry, 297, 298
Eastern Paralyzed Veterans Association
(EPVA), 312–315
Eastman Chemical, 257
eBay, 39, 49
e-Citi, 50–51, 61, 347
EDI (Electronic Data Interchange), 205
e-Dialog, 129
80 percent rule, 215–218
Electronic Data Interchange
(EDI), 205
Email, xiv, 5, 41
communication via, 118–129, 132–138
to customers, 261
and database marketing, 197–198
direct mail vs., 150–152
marketing B2B catalogs with, 288–290
Email name capture, 274
Employee benefits, 195–200
Employee loyalty programs, 223
Employee-recognition program, Webbased, 155–158
Empowerment (of customer service), 252
Engage Inc., 48
E-piphany, 4
Epiphany software, 190
EPVA (see Eastern Paralyzed Veterans
Association)
Equifax, 323
Equipment financing, 296–299
Essence (Lucinda Williams), 120, 122,
124, 126
E-Toys, 51–52, 61
E*TRADE, 39
Events-based marketing, 236–237
Exchanges, name, 270
Expedia, 39
Experian, 14, 175, 304, 308, 355
Expiration dates, 223
Failures, database marketing, 15–17
FAO Schwartz, 51
Federal Express, 209–210, 210, 213,
250, 253, 270
Filtering, collaborative, 280–282
Financial product offerings, 235
Financing, equipment, 296–299
Fleet Bank, 294
Focus groups, online, 276
Follow-up messages, 101–102, 288
Ford, 341
Forrester Research, 96, 278
"Forward-deployed" products, 205
Franks, Bill, 196, 197
Free services, fees vs., 189
Frequent flyer programs, 221, 223
Friends and Family (MCI), 16, 20
Fulfillment, 270
Fund-raising, nonprofit, 312–315
Furs & Station Wagons, 13
Future value model, 231–232
Gartner Group, 33
Gateway, 42, 188
GE Capital, 154
GEAppliances.com, 56
Globe and Mail, 306
GM, 341
Gold customers, xvii, 215, 221, 279
Golden Casket, 231–235, 246, 247
Goldfeder, Judd, 109, 111
Goodyear, 116
Google.com, 45
Gore, Al, 270
Great Universal Stores (GUS), 281, 292
Greetings, personal, 147
GUS (see Great Universal Stores)
Hallmark, xiii
Hallmark Gold Crown, 221
359
360
Inde x
Handler, Jeff, 168
Hard Scrabble, 13
Harris, Patty, 196
Harte-Hanks, 168, 169
Harvard Business School, xix
Hertz, 56
Hewlett-Packard, 148
Hilton, 221
Hoover, 42
Hoover.com, 56
Hotels, 203–204
Hudson's Bay Company, 63
Hughes, David, 48
Hughes, Helena, 45, 48, 85, 225, 270,
272–273
Hunter Business Direct, 109
Huth, Jonathan, 235
Hyatt Hotels, 3
IBM, 42, 188
Idealab, 51
iDine, 167, 183, 185, 222
iMarket, 227–231, 246, 247
iMarketing, 344
Importing function, 301
Incentives, customer, 260–261, 295
Information technology (IT)
departments, 4, 22, 113
InfoUSA, 154
Insurance, 195–200
Intel, 51, 205, 210
Intercitylines.com, 45
Internet, 41
customer service via the, 253–267
employee benefits on, 195–200
(See also World Wide Web)
Inventory-management example, 208–209
Iomega, 42–43, 332, 334, 336
Isuzu, 153–155, 161, 200–203, 210, 212,
213
IT departments (see Information
technology departments)
Ivory Soap, 84
JC Penney, xiii, 51
John Deere Credit, 116
Just-in-time, 205
KBKids, 51
Key, 297, 298
Key Bank, 296
Key Equipment, 317, 320
Kmart, 178, 179, 183, 185
KnowledgeBase Marketing, 158–162
Kodak, 47
Konves, Beth Ann, 173
Kraft Foods, xiii
Lands' End, 118, 258, 276–278, 292
Landsend.com, 276
Last-minute specials, 97–98, 288
Leads:
for equipment financing, 296–299
Web-based lead-management systems,
300–302
Lehman, Julie, 130
Lemke, Tom, 179
Lenscrafters, 106
Lexus, 302
Life insurance, 29–30
Lifetime value (LTV), 9–10, 63–81, 87,
313, 315, 340
automobile company example of, 64–75
computing name value from, 90–93
definition of, 64
keys to success and failure with, 80–81
loyalty program analysis of, 222
and multiple product ownership, 77–78
potential, 78–79
prospect, 79, 308–310
purpose of, 63
and recency, 171–173
and segmentation, 77
segments, calculating by, 93
and selling your marketing program, 80
and use of lifetime value data, 75–77
Live agent function, 256
Lost Highway, 120, 121, 122, 124,
126
Lottery programs, 231–235
Loyalty Effect, The (Fredrick Reichheld),
12, 29, 36, 293, 345
Loyalty Group, 116, 117
Loyalty Marketing, the Second Act (Brian
Woolf ), 345
Loyalty programs, 220–227, 231–235
building, 231–235
and customer service, 250–252
specialty store, 224–227
supermarket, 223–224
Web-based, 158–160
LTV (see Lifetime value)
Lundal, Jeff, xix
Lynne, Shelby, 128
M80, 124
Macy's, 279–280, 292, 329–331, 354
Macys.com, 279
Madati, Kay, 302
Madonna, 56
Managing customers (see Customer
management)
Market Facts, 158
Marketing:
of B2B catalog with email, 288–290
functions of, 63
Inde x
Marketing (Continued)
mass, 294
one-to-one, 165
profile, 322–336
quadrant, 175–178
viral, 102–104
on the Web, 38–39
(See also Customer relationship
marketing)
Marketing databases, 14–15, 166–173
Marketing departments, 113
Marketing segments (see Segments,
customer)
Marriott Hotels, 258
Mass marketing, 294
MasterCard, 166
MatchLogic.com, 48
McDoniel, Bruce, 12
MCI, xiii
McKenzie, Mac, 154
McKim, Bob, xix, 332, 334
McKinsey Quarterly, 33
Media selection, 234
Mercedes Benz, 302
Microsites, Web, 6–8, 123–124,
144–146
Microsoft, 40, 60
Microsoft Excel, 198
Models/modeling, 6, 325–328
Money & Brains, 13
Montgomery Ward, 270
Morgan, Roy, 234
Mothers-to-be, 178–179
MP3.com, 124, 126, 127
Msdbm, xix, 120, 123, 126, 153, 154,
200, 201, 303, 304, 332
Multibuyers:
performance measurement, 276
Web creation of, 279–280
Multiple product ownership, 77–78
My Personal Shopper, 277, 278
National Thoroughbred Racing
Association (NTRA), 129
NBP (see Next best product)
NCR, 34–35, 173
Neiman Marcus, xiii
Net present value (NPV), 67, 88, 92–93
Netperceptions.com, 280–281, 281
"New economics," 40
New York Post, 306
New York Times, 43, 306, 332
Newsletters, 104, 278
Next best product (NBP), 190–193
Nikkon, 144
No-hassle return policies, 271
Nonprofit fund-raising, 312–315
Northwestern University, xix
NPV (see Net present value)
NTRA (National Thoroughbred Racing
Association), 129
NuttyPutty, 51
NAB (see National Australian Bank)
Names, customer/prospect:
calculating value of, 85–93
email names, 95–106
exchanges of, 270, 294
and increasing retention rate,
93–106
packaged goods purchasers, 85
rentals of, 105–106
using LTV to compute value of, 90–93
National Australia Bank (NAB), 173, 183
National (car rental company), 56,
173–175
National Center for Database
Marketing, xiii, 344
National Retail Federation, 22, 36, 279
National Semiconductor, 209–210, 213
Package tracking system, 250
Packaged goods, xv, 85
Panduit, 208–209, 209
Parts-ordering system, 261, 264–266
Payless Auto Rental, 56
PCs, xiii, 40
Peapod, 39, 43
Penetration analysis, 295–296
Peppers, Don, 43
Performance, measuring multibuyer,
276
Personal greetings, 147
Personal profiles, 5
Personalized catalogs, 272
Personalized communications, 275
Personalized Web sites, 256
Phone matching, reverse, 167–173
O'Connor, Terri, 171, 172
Offers, timeliness and relevance of, 24–25
Office Depot, 3, 280, 292
Office Depot Online, 280
OgilvyOne Worldwide, 117
One to One Future, The (Don Peppers
and Martha Rogers), 43
One-click ordering, 275
One-to-one marketing, 165
Online exchange, 195–200
Online focus groups, 276
"Openability," 306
Operational databases, 13–15
Opt In/Opt Out, 133–136, 274
Oracle, xiv, 170
Orbitz.com, 48
Ordering costs, 274
Ordering via the Web, 42–43, 274–276,
339–340
Out-of-stock information, 282
361
362
Inde x
Photographs (in marketing), 200
Pools & Patios, 13
Postcard Direct, 200–203
Potential lifetime value, 78–79
Premier Pages, 256–258
Premiums, 45–47
PreVision Marketing, 130–132
Price, focus on, 16–17
Priceline.com, 39
PriceWaterhouseCoopers, 80
Pricing, straddle, 224
Processing function, 301
Procter & Gamble, 84
Product, focus on, 25
Product analysis, 313, 315
Product information, 289
Product managers, 25–26
Profiles, 286–288, 294, 322–336
customer-generated, 328–336
identification of, 313
identifying marketing segments with,
323
information derived from, 322–325
and modeling, 325–328
plan for, 314–315
Profitability, xvii
Promotion:
of B2B catalog with email, 288
of loyalty programs, 227
Promotion plans, 223
Promotions, 96–97
Prospect lifetime value, 79, 308–310
Prospects:
building relationship with,
302–305
identifying, 188–190, 193–195,
297–299
scoring system for, 304–305
Purchases, past, 259, 260
Purchasing officers, 260–261
Pure-play merchants, 39, 48
Quadrant marketing, 175–178
Quizzes, Web, 332–336
Radisson Hotels, 203–204, 210, 212
Ralston Purina, 136
Rapolla, Joe, 120, 121
RCI, 310, 311
Recency, frequency, monetary (RFM)
analysis, xvii, 115–117, 170
Recker, Gail, 203
Recorded music industry, 119–128
Referrals, 295, 315–317
Registration, customer, 294, 300, 334
Regression analysis, 11
Reichheld, Fredrick, xix, 12, 29, 32, 36,
293
Relational databases, xiii–xiv, 342
Relationship building:
business-to-business, 109–110
consumer, 110–111
Relationship marketing, 253
Relevance (of offers), 24–25
Repeat business, 168–169
Reporting function, 301
Reports, customer information, 260
Restraint (in customer management),
215–216
Retail Strategy Institute, Inc., 293
Retention messages, 98–101, 288
Retention rate, 10–11, 93–106
Return on investment (ROI), 26, 131
Return on promotion (ROP), 326–327
Return policies, 271
Revenio, 229
Revenue analysis (see Risk/revenue
analysis)
Reverse phone matching, 167–173
RFM analysis (see Recency, frequency,
monetary analysis)
Rhodes, Richard, 281
Ricardo, Fernando, 173
Risk/revenue analysis, 12, 218–220
R.L. Polk, 154
Rogers, Martha, 43
ROI (see Return on investment)
ROP (see Return on promotion)
Roy Morgan (research survey), 234
Ryder Truck, 154
SalesLogix, 318, 320
Schlaphoff, Evelyn, xix
Scoring system, prospects, 304–305
ScotiaApplause, 155, 156, 157, 161,
162
Scotiabank, 156, 157, 235–238, 246
Search boxes, 147–148
Sears, 42, 270
Sears Canada, 273
Secret Double Agent, 203–204
Segment marketing, 165, 182
Segments, customer, xv, 5, 9, 17, 24,
164–184, 297–298, 313, 314, 341
and automatic trigger marketing,
228–229
calculating lifetime value by, 93
creating, 170
and customer management, 235–239
database for marketing to, 166–173
guidelines for identifying and
utilizing, 182
and lifetime value, 77
maintaining, 170–173
mothers-to-be, 178–179
profiles used for identification of, 323
and quadrant marketing, 175–178
Inde x
Segments, customer (Continued)
strategic marketing decisions based
on, 242–245
targeting, 173–175
viability of, 180–182
Sequoia Capital, 51
Service:
focus on price vs., 16–17
free vs. fee-based, 189,
7-11, 44
Shah, Sejal, 285–290
Shar Van Boskirk, 96
Sharper Image, 279, 292
Shaw, Stephen, 164
Shell, 116
Sherman, Mike, 33
Shipping partnerships, 209–210
Shop.org, 279
Shotguns & Pickups, 13
SIC codes (see Standard Industrial
Classification codes)
SIGMA, 296–298
SmartDM, 196, 197, 198
Smarterkids, 51
Spare parts, 264–266
Special Olympics, 240
Sports Authority, The, 167, 168, 169,
170
SQL Server, xiv
Standard Industrial Classification (SIC)
codes, 5, 13, 191–193, 295
Staples, 3, 54, 148
Status levels, loyalty program, 223
Straddle pricing, 224
Strategic Database Marketing (Arthur
Hughes), 116, 258, 260, 345
Stride Rite Shoes, 130–132
Success:
key to, xvi
measuring, 299
Summit Marketing, 12, 326
Supermarkets:
loyalty programs for, 223–224
Web, 39, 43–44
Suppliers, 187–188
Supply chain management, 204–208
Surveys, 5, 328–332
Sweepstakes, 203
Symcor Direct Response, 117
Target, 51
Target Marketing, 344
Taschek, John, 21
Telephone calls, 5
Telesales, 271
Test groups, 252
Testing databases, 15
Tests, email, 132–138
Third-class mail, 5
Three-click rule, 148, 259
Time Warner, 40
Timeliness (of offers), 24–25
Timeshares, 310–312
Toll-free numbers, 5, 152
Top ten prospects, 188–190, 193–195
Tour selling, 311
Toysmart, 51
Trans Union, 323
Travel agents, 203–204
Travel sites, 39
Travelocity, 39
Travis, John, 63
Tupperware, 249, 258, 267
UBN (see Union Bank of Norway)
UCEA, 145
UMG (see Universal Music Group)
Union Bank of Norway (UBN), 238–239,
246, 248
United Airlines, 221
United Grain Growers, 116
Universal Music, 96, 276
Universal Music Group (UMG),
119–121, 123, 124, 126–128
University of Toronto, 170, 183
Upgrading (of donors), 313
UPS, 55, 270
U.S. Bureau of the Census, 322
U.S. Postal Service, 323
US Airways, 221
VAR Business, 134
Vendor-managed inventory (VMI),
204–209
Verizon, 111, 112, 115, 140
Verizon Database Marketing, 111,
112
Verizon Information Services, 111
Victoria's Secret, 96
Viral marketing, 102–104, 289
Virtual models, 277
Visa, 166
VMI (see Vendor-managed inventory)
Wallin, Maria, 117
Wal-Mart, 48, 51
Wang, Paul, xix
Washington Post, 43
Web Grocer, 39
Web microsites, 6–8, 123–124, 144–146
Web newsletters, 278
Web personalization, 276
Web quizzes, 332–336
Web response, 142–152, 155–161
advantages of, 142
cost savings with, 146–147
crucial elements of, 147–148
direct-mail response vs., 149–152
363
364
Inde x
Web response (Continued)
employee recognition program,
155–158
generating, 149–152
loyalty program example, 158–160
microsites for creating, 144–146
process of, 143–144
Web sites:
measuring the value of, 57–59
reasons for catalog, 274–276
Web Van, 39
Weekly Standard, 305–306, 318, 321
Weldon, 295
White mail, 152
Why They Kill (Richard Rhodes), 281
Williams, Lucinda, 120–128
Women.com, 43
Woolf, Brian, 227, 293
World Wide Web, xiv, 38–60
advertising on, 43
benefits microsite on, 222
business-to-business transactions via,
52–53
combining databases with, 6–7
as consumer research and transaction
tool, 44–45
custom promoting on, 48–49
World Wide Web (Continued)
and customer involvement, 249–250
customer service via the, 253–267
disillusionment with selling on, 40–41
examples of failed ventures on, 50–52
finding dealers via, 49–50
and growth of database marketing, 4–6
initial approach to marketing on, 38–39
keys to success for selling on, 53–57
managing leads on, 300–302
and "new economics," 40
ordering via the, 42–43, 274–276
as passive medium, 41–42
premiums/certificates distributed via,
45–47
problems with sites on, 278–279, 282
pure-play selling on, 48
"pure-play" vs. "bricks and click"
merchants on, 39
supermarkets, Web, 43–44
Yahoo!, 43
Yellow Pages, 44, 45, 111, 112, 114, 115
YourSport, 241–245
ZanyBrainy, 51
Zapdata.com, 227
Zip drive (Iomega), 42–43
About the Author
Arthur M. Hughes is vice president for business development of CSC
Advanced Database Solutions (www.cscads.com), which builds and maintains databases for major U.S. corporations. A pioneer in the use of databases to reach customers and impact their decision making, Hughes
wrote the classic marketers’ guidebooks Strategic Database Marketing
and The Complete Database Marketer. He can be reached at
ahughes@cscads.com

The Customer Loyalty Solution

Transcription

Similar documents

Worldwide Local Flyer yb 7

An open letter to Jim Ratcliffe, Ineos` owner

48 Low-Cost Business Ideas

Helping Your Family Save Money | Kids.gov

RD300 - RotoMetrics

CALLING ALCOVA MORTGAGE BROKERS

Thrislington Cubicles - design management europe award

Vladimir Luxuria Syracuse University in Florence

DOMAINfest 2015 - 2015-09-17 Xinhua

JfMUu? Jpiik Mt* Wvid 4 M{e. LOYALTY —TO YOURSELF? RESELL